The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Federal Reserve Bank of Chicago How Did Schooling Laws Improve Long-Term Health and Lower Mortality? Bhashkar Mazumder REVISED January 24, 2007 WP 2006-23 How Did Schooling Laws Improve Long-Term Health and Lower Mortality? Bhashkar Mazumder Federal Reserve Bank of Chicago January 24, 2007 Abstract Recent evidence using compulsory schooling laws as instruments for education suggests that education has a causal e¤ect on mortality (Lleras-Muney, 2005). However, little is known about how exactly education a¤ects health. This paper uses compulsory schooling laws to try to identify how education impacts health and to indirectly assess the merit of using these laws to infer the causal e¤ect of education on health. I …nd that previous Census mortality results are not robust to the inclusion of statespeci…c time trends but that robust e¤ects of education on general health status can be identi…ed using individual level data in the SIPP. However, the pattern of e¤ects for speci…c health conditions in the SIPP appears to depart markedly from prominent theories of how education should a¤ect health. I also …nd that vaccination against smallpox for school age children may account for some of the improvement in health and its association with education. These results raise concerns about using early century compulsory schooling laws to identify the causal e¤ects of education on health. I thank Douglas Almond for many helpful discussions and for contributing to some of the ideas in this paper. I also thank participants at the 2006 NBER Spring Health meetings for many helpful comments. I also thank Adriana Lleras-Muney for sharing her computer code, data and for helpful discussions. Similarly I thank Claudia Goldin for sharing her data. The views expressed here do not necessarily re‡ect those of the Federal Reserve system. 1 1 Introduction Social scientists have long been aware that there is a strong association between education levels and health (Kitagawa and Hauser 1973) but much less is known about how these factors are connected, and whether the relationship is in fact, causal. As Richard Suzman of the National Institute on Aging recently stated, "Education ...is a particularly powerful factor in both life expectancy and health expectancy, though truthfully, we’ not quite re sure why." (Lyman 2006). A recent study by Lleras-Muney (2005) provides perhaps the strongest evidence that education has a causal e¤ect on health. Utilizing a more compelling research design than most previous work, Lleras-Muney uses state compulsory school and child labor laws as instruments and …nds that increased schooling for those born in the …rst quarter of the twentieth century, led to dramatic reductions in mortality rates during the 1960s and 1970s. Her IV point estimates imply that an additional year of schooling reduces mortality risk by between 30 and 60 percent.1 These results were prominently featured in a front page story in the New York Times entitled "A Surprising Secret to a Long Life: Stay in School" (Kolata 2007) The mechanism by which schooling improves health, however, remains illusive. A variety of theories have been proposed to explain how schooling might improve health. These theories emphasize the role of eduction in a¤ecting various proximate determinants of health. Health determinants leveraged by education include: (i) …nancial resources, (ii) decision making ability, (iii) time preference. Lleras-Muney (2005) found that adjustment for income or occupation did not alter her IV estimates, and therefore discounted the role of resources behind her …ndings. Instead, her results appeared more consistent with a role of “critical thinking skills." Such skills may allow one to utilize advances in medical technology (e.g., Glied and Lleras-Muney (2003)) or manage chronic conditions better (e.g. Goldman and 1 Lleras-Muney uses the Census to estimate ten year mortality rates for synthetic cohorts using state of birth, year of birth and sex. The mean death rate in Lleras-Muney’ sample is about 10 percent and her IV s estimates are as high as -0.06 implying that a 1 year increase in schooling would lower the death rate by as much as 6 percentage points. 2 Smith (2002)). These hypothesized mechanisms both reference the ability to obtain and process information, which may be improved through education.2 In this paper I examine whether compulsory schooling laws can provide insights into exactly how education may improve health. As part of the analysis I also reassess whether compulsory schooling laws can be used to draw inferences about the causal e¤ects of education on health. As is well known, there is no test of instrument exogeneity in the exactly identi…ed case.3 I therefore conduct some (necessarily indirect) exercises to explore the validity of compulsory schooling law instrument. I begin by revisiting the mortality results in Lleras-Muney (2005) by adding signi…cantly more data (e.g. doubling the 1970 Census sample and quintupling the 1980 sample) and employing several robustness checks. The key …nding is that the Census mortality results are not robust to the inclusion of state-speci…c time trends. This raises the concern that the instruments might be picking up smooth cohort trends in educational attainment rather than discrete increases induced by more stringent compulsory schooling laws.4 I also …nd that the e¤ects of education on mortality appear to be driven primarily by the earliest cohorts (born 1901-1912) during the 1960 to 1970 period. In a second exercise I use a new microdataset: the Survey of Income and Program Participation (SIPP). The SIPP directly queries the health of each respondent, while the Census must be aggregated by synthetic cohort (by state of birth, sex and year of birth) in order to impute mortality by using estimates of "missing" individuals. The latter estimates arguably could be due in part, to a selection e¤ect if less educated individuals are also less likely to be captured by the Census over time. In contrast to the synthetic cohort approach, with the SIPP we can be sure that those who were “treated" by the compulsory school laws are indeed the same individuals registering the change in health. Moreover, as will be 2 In addition Lleras-Muney could not rule out possibility that education lowered individual discount rates, and thereby lead to healthier behaviors. See (Grossman 2005). 3 And 4I the IV estimator is inconsistent when the instrument is endogenous. also discovered a coding error that generated some erroneous OLS and IV estimates in Lleras-Muney (2005) that are quantitatively meaningful. I discuss this in more detail in section 4 3 shown later, the SIPP microdata provides relatively strong statistical power in assessing the relationship between education and health. Using the SIPP with the same IV strategy, I …nd large and statistically signi…cant e¤ects of education on health that are robust to the inclusion of state-speci…c time trends. In particular, I document that the summary health measure from the SIPP – self-reported health – also improves with changes in schooling (as induced by compulsory school laws). So while there may be some uncertainty about the robustness of the Census mortality results based on group level data, the evidence is quite strong with the individual level data using health status as an outcome. This addresses the concern about whether the health e¤ects were due to idiosyncratic changes in the laws. I then use the SIPP to examine a broad range of health measures in order to isolate which speci…c health conditions responded to education improvements induced by compulsory schooling laws. This is potentially useful for understanding whether or not the use of compulsory schooling laws as instruments produce sensible results that accord with the leading hypothesized mechanisms for how education a¤ects health. If for example, all of the health e¤ects are concentrated in only one or two health conditions that are unrelated to improvements in medical technology or decision making ability, it might cast some doubt on the validity of the instruments. If on the other hand, we were to assume that the instruments were valid, the results ought to be informative about the critical question of how exactly higher education levels lead to better health. In fact, I …nd that among the nineteen health conditions examined, only four show signi…cant declines in incidence due to education. What is striking is the absence of e¤ects among the many health conditions where decision-making ability is believed pivotal. For example, no e¤ect is found for chronic diseases such as arthritis, cancer, heart disease, lung disease, or stroke incidence. The sole exception is diabetes, where the ability to maintain a treatment regime is especially important. Moreover, education is found to increase the likelihood of hypertension and kidney problems: conditions for which self-management and 4 recent technological advance appear to be important. The lack of any e¤ects across most outcomes also suggests that channel underlying the connection between education and health is probably not due to …nancial resources or unobserved time preferences which would tend to improve health across the board. What then accounts for the positive relationship between schooling and self-reported health using these instruments? Surprisingly, health conditions where decision-making appears comparatively unimportant underly the relationship. Sensory functions –in particular hearing and vision –exhibit large and signi…cant impacts when using compulsory school laws as instruments. I also …nd that education reduces sti¤ness or deformity of the limbs, back problems, senility, and improves the ability to speak in the IV speci…cations. This pattern of e¤ects suggests that either: (i) the mechanisms by which schooling impacts health depart markedly from those hypothesized, or (ii) the use of compulsory school laws as an instrument may be suspect. An important caveat is that that with the SIPP I am using a sample of individuals who have survived into their later years (between the ages of 59 and 83) where presumably there has already been considerable positive selection on education and health. Of course among this more selected sample we might expect there to be a bias against detecting any health e¤ects, so the …nding of a strong e¤ect on overall health status might be considered surprising. Nonetheless the age of the sample raises questions about the extent to which these results generalize to the broader population. Finally, I hypothesize that schooling law changes may be correlated with other contemporaneous policies either inside or outside of schools that improved long-term health. During the early period of the twentieth century there were fairly dramatic improvements in public health measures and large declines in concurrent mortality.5 There was also a recognition that compulsory schooling was useless if students were mentally or phsycially un…t to attend school. 5 Cutler This led to other reforms in the schools that were designed to and Miller (2004), for example argue that the introduction of clean water technologies early in the century can account for half of the reduction in mortality in large cities during that time. 5 improve children’ wellbeing. In a third exercise I examine one potential factor that might s account for the observed relationship between schooling laws and improved health, smallpox vaccination. The vaccination of children against smallpox as a requirement for school entry is likely to be correlated with years of education and plausibly exerts e¤ects on adult health outcomes. Data on smallpox incidence and vaccination rates are thin, preventing de…nitive conclusions. Nevertheless, I …nd that states that appeared to have more stringent vaccination requirements for school entry experienced most of the gains in long-term health generated by compulsory schooling laws. The fact that survivors of smallpox are known to su¤er from compromised vision, hearing and speaking provides some additional suggestive evidence of a possible link between vaccination requirements and the estimated long-term health e¤ects of compulsory schooling laws. The remainder of the paper is organized as follows. In section 2 I review the relevant literature, in section 3 the Census and SIPP data are described the econometric models are shown, in section 4 the baseline results are presented, in section 5 I consider the possible role of smallpox and vaccination in schools and in section 6 I conclude. 2 Literature Review It has been over thirty years since Grossman published his seminal economic model of health determination (Grossman 1972). This model includes the assumption that education increases the e¢ ciency of health production. And while Grossman’ conceptual framework s has served as the “work horse" model for applied work in health economics, little is understood about how or what kinds of education enable the production of health.6 For example, in 2003, the National Institutes of Health solicited (quite general) research proposals on the “Pathways Linking Education to Health." This RFA sought “validation of speci…c measures of abilities crucial to educational attain6 Grossman (2005) noted that “extensive reviews of the literature [concluded that] that years of formal schooling completed is the most important correlate of good health. 6 ment, such as level of cognitive or language skills" that improved health, and even cautioned that: “The association or pathway between formal education and either important health behaviors or diseases may not be causal. Instead it may re‡ the in‡ ect uence of confounding or co-existing determinants or may be bi-directional."(NIH 2003) Recent years have witnessed an upsurge of interest in education’ role in determining s health. In one widely-cited paper, Goldman and Smith (2002) noted that more educated patients may manage chronic conditions better. Those with more schooling adhere more closely to treatment regimens for HIV infection and diabetes, which can be fairly complex. For such conditions, the ability to form independent judgements and comprehend treatments is important, and apparently is fostered by schooling. Accordingly, “self-maintenance is an important reason for the very steep SES gradient in health outcomes" (Goldman and Smith (2002):10934). Glied and Lleras-Muney (2003) looked at health conditions experiencing more rapid technological change, …nding that more educated respondents faired better. They argued that “the most educated make the best initial use of new information about di¤erent aspects of health" permitting them to respond more adeptly to evolving medical technologies. They noted that no consensus measure existed for assessing the pace of innovation in health. They therefore consider several measures, including the change in mortality rates for speci…c conditions from 1986 to 1995 and the number of patents issued for particular conditions. They found that education gradients were steeper for diseases that were more innovative by these measures. A growing literature has also tried to examine whether the education gradient in health is causal by using instrumental variables. Reviews of these studies may be found in (LlerasMuney 2005) and (Grossman 2005). While these studies typically …nd an e¤ect of more education leading to better health, in most cases it is questionable whether the instruments are truly exogenous.7 In contrast, the use of changes in compulsory schooling laws appears 7 For example Leigh and Dhir (1997) use parent schooling, parent income and state of residence as instruments, all of which could plausibly a¤ect long-term health independently of their e¤ects through 7 to be a more compelling instrument choice since it is more plausibly exogenous than instruments used in prior work. Previous studies also typically have looked at just one or two health outcomes and have not systematically compared the e¤ects across a range of health outcomes to distinguish between competing theories of how education a¤ects health. 3 Data and Methodology 3.1 Mortality Data and Econometric Speci…cation I begin by describing the procedure used to estimate the e¤ects of education on mortality in Lleras-Muney (2005). This will provide the basic framework for extending the analysis to examining other health outcomes in the SIPP and for expanding the analysis along other dimensions. I brie‡ describe the approach here, for a more detailed discussion that y includes alternative estimation strategies see Lleras-Muney (2005).8 The key idea is that in the absence of a large sample tracking individuals over their entire lifetime, synthetic cohorts are constructed using Census data. With the Census, we know age, completed education and state of birth which allows us to infer the compulsory schooling laws that a¤ects each cohort in each state of birth. Mortality can be measured by tracking population counts of particular groups across Census years. The mortality rate at time t for cohort c, of gender, g, born in state s, (Mcgst ) is simply measured as the percentage decline in the population count (Ncgst ) within these cells over the subsequent ten years: Mcgst = Ncgst Ncgst+1 Ncgst (1) schooling. 8 Lleras-Muney also uses several other approaches. She estimates the model at the individual level using data from the National Health and Nutrition Examination Survey (NHNES). This is largely for comparative purposes since the sample is too small to estimate statistically signi…cant e¤ects using IV. She also considers Wald estimators and introduces a a "mixed" two stage least squares approach using individual data in the …rst stage but aggregate data in the second satge. The latter two approaches produce roughly similar estimates to the aggregate IV estimates which are modeled here. 8 The mortality rate for each cell is then modeled as follows: Mcgst = a + Ecgst + Wcs + c + s + cr + t + d + "cgst (2) where Ecgst is the average education level for that cell at time t, Wcs , measures a set of cohort and state speci…c controls measured at age 14 intended to capture di¤erences in other potential early life determinants of mortality (e.g. manufacturing share of employment, doctors per capita). The model also includes a set of cohort dummies dummies s, interactions between cohort and region of birth, year dummies, cr ;a c, state of birth female dummy d and t. I construct two datasets for the analysis. The …rst attempts to replicate Lleras-Muney point estimates and uses the same 1 percent ipums samples drawn from the 1960, 1970 and 1980 Censuses that are produced by the Minnesota Population Center.9 The second estimation sample replaces the 1970 1 percent sample with a 2 percent sample that combines the form1 and form 2 "state" 1 percent samples.10 For 1980 I use a 5 percent sample. In addition, I also add the 5 percent samples for 1990 and 2000. All of the Censuses are scaled appropriately to produce population estimates that correspond to Ncgst 11 : Following Lleras-Muney I restrict the anlaysis to cohorts born between 1901 and 1925. I also follow her sample restrictions to exclude immigrants, blacks, and to topcode years of education at 18 starting in 1980. For the expanded samples I also exclude cases where age, state of birth and education are imputed by the Census Bureau. The descriptive statistics for both samples are shown in Table 1. It is worth noting that the death rate for the 1970 to 9 Ruggles et al. [2004]. 1 0 Unfortunately, combining any of the other four 1 percent samples that are available for 1970 would lead to geographically unrepresentative samples. 1 1 The 1960, 1970 and 1980 samples are self-weighting samples so the raw population counts can simply be scaled up by multiplying by 100, 50, and 20. representative estimates of the population. The 1990 and 2000 samples require weights to produce We found that using the self-weighting 1 percent samples for 1990 and 2000 instead of the 5 percent samples had little e¤ect on the point estimates but increased the standard errors. 9 1980 period is quite a bit larger with the expanded sample but that the standard deviation is about 20 percent lower. There are now also 5 additional cells that had missing data when using just the 1 percent samples. The death rates for the 1980-90 and 1990-2000 periods are much higher due to the fact that I follow these same cohorts when they are much older. Figure 1 plots the death rates by age for each Census year. This highlights the importance of controlling for age in the speci…cations which is done by adding polynomials in age to the models. One straightforward way to estimate in (2) would be through weighted least squares (WLS), with the weights corresponding to the population represented by each cell. However, that this would produce a biased estimate due to omitted variables. Any number of factors could plausibly be associated with both higher education and lower mortality even at the group level. Therefore, two stage least squares is used where in the …rst stage education is instrumented with the set of compulsory schooling laws, CLcs , in place for each cohort and state of birth: Ecgst = b + CLcs + Xcgst iv + Wcs iv + iv;c + iv;s + iv;cr + t + ucgst (3) The instruments for the compulsory schooling laws are constructed in the following way. The variable childcom measures the minimum required age for work minus the maximum age required for a child to enter school, by state of birth and by the year the cohort is age 14. This variable takes on one of eight values (that range from 0 to 10). A set of indicator variables (excluding the 0 category) are used as instruments. In addition there is an indicator for whether school continuation laws were in place in that state. These laws required workers of school age to continue school part-time. For comparability, I use the same dataset as Lleras-Muney (2005).12 In addition to a more detailed description of these variables in Lleras-Muney (2005), there is also an extended discussion of these measures and their appropriateness as instruments in Lleras-Muney (2001). The variables contained in 1 2 Downloaded from Lleras-Muney’ website. s 10 Wcs are the corrected versions of those used by Lleras-Muney (2005). I also experiment with a second set of data independently collected by Goldin and Katz (Goldin and Katz 2003). Goldin and Katz carefully compared their series with the LlerasMuney data and the data collected by Acemoglu and Angrist (2000) and recti…ed di¤erences wherever possible. Since the Goldin and Katz data go back further in time it is possible to match all of the cohorts to the school entry age laws in e¤ect when the cohorts were younger than 14. I use this data to measure the required age for school entry when the cohorts were likely to be at age 8 instead of age 14. In principle, incorporating this data should provide a better measure of the total years of compulsory schooling. 3.2 Health Microdata and Speci…cations The second sample is constructed by pooling individuals across various panels of the Survey of Income and Program Participation (SIPP) during the 1980s and 1990s.13 Because participation in many programs is closely related to an individual’ health and disability status, s the SIPP routinely collects information on health and medical conditions. The SIPP is also ideally suited for this analysis because it contains the state of birth of all sample members allowing us to implement the IV strategy of using compulsory schooling laws during childhood. One useful outcome is self-reported health (SRH). SRH is on a 1 to 5 scale where 1 is “excellent” 2 is "very good", 3 is "good", 4 is "fair" and 5 is “poor” SRH has been , . found to be an excellent predictor of mortality and changes in functional abilities among the elderly.(Case, Lubotsky, and Paxson 2002). I experiment with this measure in a few ways. First it is simply used as a continuous variable. Second, I use indicators for being in poor health or, fair or poor health. Finally I use the health utility scale that scales the di¤erences between the categories based on a health model using the National Health Interview Survey.14 1 3 This 1 4 See includes the 1984, 1986-1988, 1990-1993 and 1996 panels. (Johnson and Schoeni 2005) and the citations there for a discussion of this approach. 11 A few other general outcomes are also examined. These include whether the individual was hospitalized during the past year, the number of times she was hospitalized, the total number of nights spent in the hospital and the number of days spent in bed in the last four months. There are also a set of questions dealing with functional activities, activities of daily living and instrumental activities of daily living.15 I assembled a common set of questions that were consistently asked across surveys. These include whether the individual has “di¢ culty" with seeing, hearing, speech, lifting, walking, climbing stairs and, whether the person can perform any of these activities "at all". In addition there is information on whether individuals have di¢ culty getting around inside the house, going outside of the house or getting in or out of bed, and whether they need the assistance of others for these activities. For a subset of individuals who report limited abilities in certain tasks or who have been classi…ed as having a work disability, detailed information is collected on a number of very speci…c health conditions including: arthritis or rheumatism; back or spine problems; blindness or vision problems; cancer; deafness or serious trouble hearing; diabetes; heart trouble; hernia; high blood pressure (hypertension); kidney stones or chronic kidney trouble; mental illness; missing limbs; lung problems; paralysis; senility/dementia/Alzheimer’ disease; s sti¤ness or deformity of limbs; stomach trouble; stroke; thyroid trouble or goiter; tumors (cyst or growth); or other.16 Since the speci…c health ailments are only asked of speci…c subsamples, they probably only pick up on the most severe cases. Even though, many of 1 5 These measures are derived from speci…c codes of the International Classi…cation of Impairments, Dis- abilities and Handicaps (ICIDH). 16 I pool responses from the 1984, 1990-93 and 1996 SIPPs in order to maximize sample size. Unfortunately di¤erent criteria were used across the SIPP survey years to select the subsamples for which speci…c health conditions were asked. For example, in 1996 the health conditions were asked of those who reported being in fair or poor health. I found that it was important to combine all of the subsamples in all of the years in order to have enough power to identify e¤ects. There are also an additional set of 10 outcomes that are not used because they were not available in the 1984 SIPP. Experimentation with a smaller sample suggests that the conclusions are not altered by dropping these other outcomes 12 our sample individuals are not actually asked about these speci…c health conditions we still include them in the estimation sample so that our sample is not a selected sample of only those in poor health. The summary statistics for this data are shown in Table 2. Since most of the outcomes in the SIPP are indicator variables I now use Two Stage Conditional Maximum Likelihood or 2SCML (Rivers and Vuong 1988) rather than IV.17 Rivers and Vuong show that 2SCML has desirable statistical properties, is easy to implement and produces a simple test for exogeneity. I continue to use IV for the few continuous dependent variables. Also all of the analysis is now done using individual level data. The statistical model is similar to (2) only now I use the latent variable framework: yit = a + Ei + Xi + Wcs + c + s + trends + t + "it yit = 1 if yit > 0; yit = 0 if yit < 0 (4) (5) In the …rst stage, I run a similar regression as before: Ei = b + CLcs + Xi + Wcs + c + s + trends + t + "it (6) To implement 2SCML, I use the predicted residuals from (6), bit ; and include it as an " additional right hand side variable (along with the actual value of Ei ) when running the second stage probit. For comparability I use the same sample restrictions and covariates as Lleras-Muney with only a few exceptions. Unlike Lleras-Muney I include a quadratic in age. In addition state-speci…c cohort trends are used to address concerns that region of birth interacted with cohort may not adequately control for state-speci…c factors that are smoothly changing over time.18 17 I thank Jay Bhattacharya for this suggestion. In a previous version of the paper I found very similar results using two stage least squares for the dichtomous outcomes. 18 I generally found that the IV results were larger and more signi…cant when using the state trends than when using region of birth interacted with cohort. The OLS results were virtually identical under either 13 4 Baseline Results 4.1 Mortality This section begins by describing the replication of Lleras-Muney (2005) and the discovery of a coding error concerning the appropriate base period. Using the correct base period increases the point estimates: the new WLS estimate is approximately doubled and the IV estimate is 50% higher than reported in Lleras-Muney (2005). I then expand the Census data by: (i) expanding beyond the 1% sample to the 2% and 5% IPUMS samples available for 1970 and 1980, respectively, and (ii) incorporate the 1990 and 2000 Census data in the mortality analysis. 4.1.1 Replication and Correction For the mortality analysis I start with the same sample as Lleras-Muney (2005) I have an identical number of individuals (814,805) drawn from the 1960 and 1970 Census and match nearly all the summary statistics in her Table 1 Nevertheless I …nd some large di¤erences when implementing WLS or IV at the cell level. This shown in Table 3. The …rst two columns show the WLS and IV estimates from Lleras-Muney while columns 3 and 4 show my estimates. Compared to her WLS estimate of -.017, I obtain a coe¢ cient of -.036. The di¤erence in the estimates is statistically signi…cant. For IV, once again I obtain a much larger estimate (-.072) than her estimate of (-.051). After some experimentation I speculated that Lleras-Muney did not use education calculated in the base period, t, but instead calculated education in period t+1. If I use education calculated in t + 1, then my estimates are much closer to hers. These are shown in columns 5 and 6. After graciously providing her computer code I con…rmed with Lleras-Muney that this was indeed the case and she has since written an errata (Lleras-Muney 2006) providing corrected estimates.19 The problem with using education in the later period is that the speci…cation. 1 9 The errata (Lleras-Muney 2006) reports an IV coe¢ cient of -0.063 (0.024) compared to the estimate of 14 sample has already experienced selective mortality based on education. The discrepancy between the results are not only statistically meaningful but are quantitatively important. Taken at face value, the "corrected" IV result implies that an extra year of schooling reduces the likelihood of dying over the next ten years by more than 7 percentage points. In the sample, the mean death rate is only about 10.6 percent. This suggests that one more year of schooling lowers mortality risk by nearly 70 percent – result a that is perhaps implausibly large. 4.1.2 Expansion of the Census Sample In Table 4, I show how the results change as the sample is enlarged and the speci…cations are modi…ed. I begin by just focusing attention on the …rst two columns showing the WLS and IV estimates. In panel A I isolate the e¤ects of using larger samples for 1970 and 1980. Row 1 repeats the results from Table 3. In row 2 I …nd that the WLS estimate rises to -0.045 and that the IV estimates drop considerably to -0.043. The greater precision is evident in the standard error for the IV estimate which declines by about 25 percent. In rows 3 and 4 I control for age and …nd that this lowers the WLS estimates a little and increases the IV estimates a little. In row 5 I drop the region of birth interactions with cohort and instead use state speci…c linear (cohort) trends. This raises the WLS estimate but I now …nd that IV coe¢ cient is signi…cantly lower and is no longer statistically signi…cant at conventional levels. In panel B I add data from the 5 percent samples of the 1990 and 2000 Censuses. With this larger dataset I construct death rates over four ten year periods and therefore follow cohorts over a longer period of time with a considerably larger sample. Given that the sample also tracks the cohorts later in life when mortality rates are much higher, the age controls are essential. I use a cubic in age although I …nd that the results are not very sensitive to the choice of the polynomial. Since medical technology and other health-related factors might change over time, I have also interacted the cubic in age with the Census -0.072 reported here. I was unable to resolve this di¤erence. 15 year. In my preferred speci…cation (row 6) I now …nd that both the WLS and IV estimates are about -0.035 which appear to be much more plausible. With this larger sample the inclusion of state speci…c cohort trends again results in a point estimate that is much smaller in magnitude (-0.02) and not statistically distinguishable from zero (row 7). The third column shows the e¤ects of using the Goldin and Katz data for constructing the instruments. For most speci…cations in panels A and B they produce similar estimates as the baseline IV results although the standard errors are a bit higher. It is worth noting that with the Goldin and Katz data the inclusion of state speci…c cohort trends lowers the size of the point estimates even more dramatically and also yields the same conclusion, that the estimates are not statistically di¤erent from zero. 4.1.3 E¤ects by Subgroups In the remaining panels of Table 4 I examine how the e¤ects vary by year, age and cohort. In panel C I separately estimate the education coe¢ cient for each Census year. Since the speci…cation includes a full set of cohort dummies these are equivalent to age controls when using a single Census year. Although the WLS estimates are signi…cant in all years they peak in 1970 at -0.061 and drop to only -0.015 by 1990. The IV estimates have large standard errors so they are likely to be imprecisely estimated. Nonetheless the estimates are large only for 1960 and are essentially zero for 1980 and 1990. This is true whether I use the Lleras-Muney data or the Goldin Katz data. In panel D I stratify the sample by three age ranges: 35-55, 55-65 and 65-89. Here I observe di¤erent patterns across the three columns making it di¢ cult to interpret the estimates. The WLS and IV estimates from column 2 suggest that the largest e¤ect may be for those aged 55 to 64, while the IV estimates with the Goldin Katz data suggest the opposite. Given the imprecision of the estimates I cannot draw any meaningful inferences regarding the age pattern Panel E however, provides a striking result that appears to be consistent across the two IV speci…cations. It appears that the entire e¤ect of education on mortality arising from 16 compulsory schooling laws is due to cohorts born from 1901 to 1912 who constitute just over 40 percent of the sample. In fact for those born from 1913 to 1925, the point estimate is actually positive in both column 2 and column 3. Using the Lleras-Muney data, the estimate is 0.035 and is signi…cant at the 7 percent level. These results taken as a whole suggest that upon closer inspection, the results from Lleras-Muney are driven by cohorts born very early in the century and their mortality experience during the 1960-1970 period. One possible explanation could be that the e¤ect of education stayed roughly constant but that compulsory schooling laws had its biggest bite for those born earlier in the century. However, I have run the …rst stage regressions by these cohort groupings and found that the partial F-statistics on the instruments are similar for both cohorts when using the LlerasMuney data and are actually higher for the 1913 to 1925 cohorts when using the Goldin Katz data. This suggests that the schooling laws may actually have been more binding for the later cohorts casting doubt on this alternate explanation. In other estimates that are not shown in the table I found no statistically signi…cant di¤erence between men and women although the point estimates were larger (in absolute value) for men using the Lleras-Muney data and very imprecisely estimated using the Goldin Katz data. 4.2 Health Outcomes from the SIPP In Table 5 the results using the microdata on health outcomes using the SIPP are presented. The …rst column shows the e¤ects of education using a simple probit (or OLS) which does not account for endogeneity. The second column presents the 2SCML (or IV) estimates using the compulsory schooling laws as instruments. Given the possible e¤ects of education on mortality and the fact that outcomes in the SIPP are not observed until at least 1984, one might not expect any remaining health e¤ects to be apparent. As it turns out I do …nd signi…cant e¤ects using the instruments for several broad health outcomes. The …rst row shows that self reported health measured as a continuous variable is a¤ected by education. The IV estimate of -0.23 is more than twice the OLS estimate (-0.09). In column 4 using 17 a Hausman test one can reject that the OLS and IV coe¢ cients are the same at the 7 percent level. Translating SRH into a health index on a 1 to 100 scale following Johnson and Schoeni’ (2005) approach, the IV estimate implies that an increase in schooling by s one year improves the health index by 4.5 points or about 7 percent evaluated at the mean (column 3). I also estimate that the probability of being in fair or poor health is reduced by 8.2 percentage points with an additional year of schooling – fairly large e¤ect that is a statistically di¤erent from the naive probit at the 18 percent sign…cance level. I do not …nd, however, that any of the measures of hospitalization or days spent in bed are signi…cant when accounting for endogeneity. Looking across a variety of measures of physical functional abilities, I …nd that while all of the naive probit estimates are sign…ciant and of the expected sign, the two stage estimates are typically not signi…cant. Given the large health e¤ects discussed above it is striking that those who have an additional year of schooling due to compulsory schooling laws are no more likely to have trouble lifting, walking, climbing stairs, getting around the house, getting around inside the house or getting into or out of bed. In fact for many of these outcomes the coe¢ cients are actually positive! On the other hand, those with greater schooling associated with compulsory schooling laws are dramatically less likely to experience problems with vision, hearing or speaking. In almost all of these cases the di¤erences between the simple probit and the 2SCML estimates are very large and statistically di¤erent at about the 10 percent level. For example, the 2SCML estimates imply that an additional year of schooling reduces the probability of having trouble "seeing" by 5.6 percentage points. In this sample the mean rate of this health outcome is 13.6 percent. These results might suggest that the channel by which general health is compromised for those with less schooling, may be related to sensory functions. The next set of results estimate the incidence of speci…c health conditions. Recall that these conditions are only identi…ed for subsets of individuals and that the screening criteria has changed across SIPP survey years. Also recall that all individuals are included regard- 18 less of whether they were screened for this question so as to avoid using a selected sample of only those in poor health. Generally, the underlying health conditions were only asked of individuals who reported particular kinds of activity limitations, reported having a work disability or reported being in fair or poor health. This is captured by the variable "di¢ culty" which, not surprisingly, is sign…ciant under both probit and 2SCML. When I turn to the estimated likelihood of having one of the underlying health conditions, the probit estimates once again are signi…cant in every case. The 2SCML estimates, however, are only negative and sign…cant for four outcomes: back or spine problems; sti¤ness or deformity of a limb; diabetes and senility/dementia/Alzheimer’ disease. It is important to point out that s "trouble seeing", "trouble hearing" and "trouble speaking" were never used as a screening criteria for asking about an underlying health condition. This likely explains why blindness and deafness are not signi…cant with the subsamples. Another interesting result is that both kidney problems and hypertension appear to positively associated with more schooling. This is especially notable because these are two outcomes for which self-management and recent technological advance appear to be especially important. According to Appendix Table B of Glied and Lleras-Muney (2003), treatment of kidney infections experienced substantial innovation. Among the 56 causes of death, it experienced the fastest decline in age-adjusted mortality from 1986 to 1995 falling at more than nine percent per year (Glied and Lleras-Muney (2003)):8, Appendix Table B). Accordingly, a steep (negative) gradient between education and kidney disease would presumably be expected. It is therefore of note that the 2SCML speci…cation …nds an increase in the incidence of kidney problems among those with high education. Treatment of diabetes is “often considered the prototype for chronic disease management." Goldman and Smith (2002). Our …ndings, which analyze a broad range of health conditions and chronic diseases, would suggest that insofar as the formal schooling is concerned, diabetes appears to be an exception. In the SIPP data, only diabetes enters enters in the expected direction –i.e. increases in schooling appear to reduce diabetes incidence. An alternate explanation 19 for the diabetes result could be that states that had higher compulsory schooling levels also promoted nutritional policies that might have reduced adult onset of diabetes. Overall, however, one conclusion that may be drawn from this table is that there is little support for the "decision-making" hypothesis. It is also worth noting that explanations for the link between education and health that focus on resources (e.g. income, occupation) or unobserved time preferences do not appear to be consistent with these results. These explanations would likely imply that all outcomes ought to be a¤ected not just a few. The major caveat to this analysis is that we observe individuals only if they have survived into the 1980s and 1990s when they are anywhere between the ages of 59 and 83 This sample is almost certainly positively selected on education and health, making it unclear how generalizable these results are. I suspect that due to this selection the results are biased against …nding any e¤ects of education on improving health, making it still surprising why there are very large negative coe¢ cients on the incidence of several negative health outcomes. 5 5.1 Smallpox Vaccination Alternative Explanations The results thus far suggest present something of a puzzle as to exactly how compulsory schooling laws early in the twentieth century led to improved long-term health status. While the results cast doubt on the traditional explanations o¤ered in the literature of how education improves health the results do not appear to point to any obvious alternative explanation. One general hypothesis worth considering is that schools served as an important place for implementing a variety of policies that may have impacted both education and health directly. It could be that states and cities during this period were introducing many reforms contemporaneously and schools were one obvious target for these reforms. 20 In fact it was noted at the time that it was pointless to force kids to attend schools if they were unable to learn. In 1904, Robert Hunter wrote in the book Poverty: "There must be thousands -very likely sixty or seventy thousand children-in New York City alone who often arrive at school hungry and un…tted to do well the work required. It is utter folly, from the point of view of learning, to have a compulsory school law which compels children, in that weak physical and mental state which results from poverty, to drag themselves to school and to sit at their desks, day in and day out, for several years, learning little or nothing." In fact in response to this situation Philadelphia, Boston, Milwaukee, New York, Cleveland, Cincinnati and St. Louis all began large scale programs to provide food in public schools during the 1900s and 1910’ (Gunderson 1971). s It seems plausible, then that coincident with the enactment of compulsory schooling laws there were likely many e¤orts (legislative or otherwise) to improve the general condition of children. In this section I pursue one speci…c alternative hypothesis that might explain some of the …ndings. Speci…cally, I examine whether the association between compulsory schooling laws and health may have been due in part, to early century school requirements concerning vaccination against smallpox. 5.2 Background on Smallpox Before Edward Jenner invented the …rst vaccine in 1797, smallpox was a widespread and brutal disease killing about 400,000 Europeans a year with survivors accounting for about one-third of all cases of blindness (Henderson and Moss 1999). Smallpox was especially concentrated among children, in the early 19th century smallpox accounted for one-third of the deaths of all children(George Palmer and Ingen 1930). More than a century after the development of the vaccine, smallpox remained a deadly threat in the United States. A report in the New England Journal of Medicine in 1930 showed that between 1919 and 1928 there were more than half a million cases of smallpox in the US and argued that ". . . the United States remains now . . . the most smallpox ridden country in the world bar possibly 21 China, India and (doubtfully) Russia.” In addition to blindness, survivors of smallpox are also known to have a higher rate of encephalitis20 (in‡ ammation of the brain). Although encephalitis is relatively rare, milder forms of the condition are likely to go unreported.21 Symptoms of encephalitis include problems with speech, hearing and double vision.21 This suggests that vaccination against smallpox in schools may have reduced the incidence of compromised sensory functions as we …nd in our SIPP sample. 5.3 Vaccination in schools States began to require vaccination against smallpox in schools beginning in the late 19th century (Hanlon 1969). I have been able to compile information concerning state laws regarding school vaccination for the years 1915, 1921, 1926 and 1941. In the …rst snapshot in 1915, fourteen states had requirements for vaccination. In the other three years I found no cases of any additional states requiring vaccination for schools. Similarly I found no cases of any states repealing these laws. Therefore, I am unable to construct an analogous panel design as employed by Lleras-Muney for compulsory schooling laws since there is no variation over time. I also assembled data on states who had laws authorizing the use of vaccination in the case of outbreaks but found that these laws relied critically on enforcement. There were also a few cases where states changed laws regarding the prohibition of vaccination in schools but it is doubtful that these law changes have enough power to identify e¤ects in the samples used in thsi study. Fortunately, a 1930 White House sponsored report on the state of young children’ s health does contain detailed data on young children’ vaccination rates by age for 156 of the s largest cities (George Palmer and Ingen 1930). For the time period this was an impressive data collection e¤ort where information was collected from around 3000 doctors and other 2 0 See AMA (1999) 2 1 See http://www.ninds.nih.gov/disorders/encephalitis_meningitis/detail_encephalitis_meningitis.htm 22 health providers on the frequency of health exams and dental exams in addition to rates of vaccination against smallpox and immunization against diptheria. The data on smallpox vaccination rates was aggregated to the state level. This is displayed in Table 6. What is striking is how the vaccination rate jumps sharply from age four to age …ve in many states as children prepare for school entry. Although some of the richer states in the Northeast like New York have sizable jumps, there is a great deal of variation even within regions. For example, Colorado, Georgia and Kentucky are among the states with largest increases in vaccination between age four and age …ve. These data also illustrate the potential pitfall of using actual state laws since some of the states that ostensibly required vaccination did not exhibit big increases (e.g. Arkansas, South Carolina) while other states that did not legally require vaccination, in practice, exhibited large increases in vaccination by school age. Although there is only vaccination rate data for one year, 1930, I use this data to test the extent to which the health e¤ects of education may be operating through di¤erences in vaccination policy. Speci…cally, I consider how di¤erent the health e¤ects of education are for states that have stringent vaccination laws versus those that don’ I assume that t the stringency of school vaccination requirements can be proxied by the change in the vaccination rate from age four to age …ve. Obviously, this approach is only ideal for the youngest cohort (born in 1925), who would have turned …ve in 1930. I analyze this …rst for the baseline mortality sample (row 6 in Table 4). The sample is split by states with a change in vaccination rate of more than 10 percentage points (the median change) and those with a change that is 10 points or fewer. The results are shown in Table 7. I …nd that the IV estimates of the e¤ects of schooling on mortality are statistically signi…cant only in the states with large increases in vaccination by school age and that the IV coe¢ cients are of the wrong sign in the states with less stringent vaccination requirements. I experimented with randomly splitting the sample 100 times and found that the odds of …nding an equivalent result by chance are only about 15 percent. I then performed the same exercise with the SIPP sample looking at the outcomes that 23 I found to be signi…cant for the full sample. In this sample the results are more mixed. I show a representative set of results from the SIPP in Table 7 For self reported health, the estimates are actually larger and more statistically signi…cant in the states with relatively less stringent vaccination requirements. However, the e¤ect on being in poor health is only apparent in the high vaccination states. Since poor health is a strong predictor of mortality this appears to be consistent with ther mortality …nding. Most of the other estimates by subsample are too noisy to say much of anything but it does appear that hearing is strongly a¤ected in the states with more stringent school vaccination requirements. Since I do not have a time series on vaccination rates and given the relative bluntness of the approach I only claim that these results are suggestive of a possible mechanism relating compulsory schooling laws and long-term health that operates through school vaccination. 6 Conclusion This paper expands upon the growing literature that attempts to identify whether there is a causal e¤ect of education on health by also considering how education might a¤ect health. I closely examine the e¤ects of education induced by compulsory schooling laws early in the twentieth century on long-term health using several approaches. First I revisit the results in Lleras-Muney (2005) by expanding the Census sample and employing a variety of robustness checks. The main …nding is that the e¤ects of education on mortality induced by changes in compulsory school laws are not robust to including state speci…c time trends. I also …nd that all of the e¤ects are for cohorts born between 1901 and 1912 and their mortality experience during the 1960s. Second, I use the SIPP to identify not only general health e¤ects but also speci…c health outcomes that were induced by changes in state compulsory schooling laws to see if these outcomes correspond to our existing theories of how education a¤ects schooling. The results suggest that there is a large e¤ect of education on general health status arising from compulsory schooling laws that is robust to state time trends. However, I …nd that with 24 the sole exception of diabetes none of the other speci…c health conditions that are associated with education (e.g. vision, hearing, speaking ability, back problems, deformities, senility) correspond to the leading theories of how education improves health (e.g. technological improvements, better decision-making, higher income). This suggests that either our theories are incorrect or that the compulsory schooling laws are suspect instruments. An impor- tant caveat, however, is that the SIPP analysis uses a sample of older individuals who are almost surely positively selected on education and health. While this likely makes it more di¢ cult to detect e¤ects of education on improved health it also raises questions as to how generalizable these results are. Third, I look at one speci…c alternate hypothesis of how state-level compulsory schooling laws might have in‡ uenced long-term health, namely through requirements for smallpox vaccination as a condition for school entry. I stratify the samples by states with stringent versus nonstringent vaccination requirements and …nd that all of the e¤ects of education on mortality and poor health status were registered in states with stringent vaccination requirements. This provides some suggestive evidence that smallpox vaccination may account for some of the link between education and health induced by compulsory schooling laws. It is also worth noting that survivors of smallpox are known to su¤er from compromised vision, speaking and hearing which are among the few e¤ects that we detected in our IV results with the SIPP. I conclude from these exercises there is reason to be concerned about whether compulsory schooling laws can be used as instruments to draw meaningful inferences about the causal e¤ects of education on long-term health. Instead it could well be that either other school-based reforms directly impacted long-term health or that other reforms with long term impacts took place at the same time that compulsory schooling requirements became more stringent. In any event, the results suggest that even if there is a causal e¤ect of education on health there is still a great deal of uncertainty about how education improves health that should remain an important topic for further research. 25 References Case, A., D. Lubotsky, and C. Paxson (2002, December). Economic status and health in childhood: The origins of the gradient. American Economic Review 92 (5), 1308. George Palmer, M. D. and P. V. Ingen (1930). Health Protection for the Preschool Child: A National Survey of the Use of Preventive Medical and Dental Service for Children Under Six. The Century Co. Glied, S. and A. Lleras-Muney (2003, May). Health inequality, education and medical innovation. NBER Working Paper No. 9738 . Goldin, C. and L. F. Katz (2003, November). Mass secondary schooling and the state. NBER Working Paper 10075 . Goldman, D. P. and J. P. Smith (2002, August). Can patient self-management help explain the ses health gradient? Proceedings of the National Academy of Sciences 99 (16), 10929– 10934. Grossman, M. (1972). On the concept of health capital and the demand for health. Journal of Political Economy 80, 223– 255. Grossman, M. (2005, August). Education and nonmarket outcomes. Mimeo, National Bureau of Economic Research. Gunderson, program. G. W. FNS (1971). 63 - History Food of and the national Nutrition school Service, lunch USDA, http://www.fns.usda.gov/cnd/Lunch/AboutLunch/ProgramHistory.htm. US GOVERNMENT PRINTING OFFICE:1971 0-429-783. Henderson, D. A. and B. Moss (1999). Vaccines, Chapter Smallpox and Vaccinia. Philadelphia: W.B. Saunders. Johnson, R. C. and R. F. Schoeni (2005, May). Early life events and health and labor market outcomes in adulthood. UC Berkeley Goldman School of Public Policy. 26 Kitagawa, E. M. and P. Hauser (1973). Di¤ erential Mortality in the United States: A Study in Socioeconomic Epidemiology. Cambridge MA: Harvard University Press. Kolata, G. (2007, January 3). A surprising secret to a long life: Stay in school. New York Times, A1. Lleras-Muney, A. (2005, January). The relationship between education and adult mortality in the united states. Review of Economic Studies 72 (1), 189– 221. Lleras-Muney, A. (2006). Errata for the relationship between education and adult mortality in the united states. Review of Economic Studies. Lyman, R. (2006, March 10). Report foresees no crisis over aging generation’ health. s New York Times, A19. NIH (2003, January). Pathways linking education http://grants1.nih.gov/grants/guide/rfa-…les/RFA-OB-03-001.html. to RFA: health. OB- 03-001. Rivers, D. and Q. H. Vuong (1988). Limited information estimators and exogeneity tests for simultaneous probit models. Journal of Econometrics 39, 347– 366. Ruggles, S., M. Sobek, T. Alexander, C. A. Fitch, R. Goeken, P. K. Hall, M. King, and C. Ronnander (2004). Integrated public use microdata series: Version 3.0. Minnesota Population Center. 27 Table 1: Summary Statistics for IPUMs samples 1960-2000 1%, 2% or 5% 1960-1980 1% only Variables Mean Ten Year death rates overall 1960-70 1970-80 1980-90 1990-00 Std. dev. N Mean Std. dev. N 0.108 0.110 0.105 --- 0.136 0.119 0.152 --- 4792 2395 2397 --- 0.213 0.113 0.154 0.287 0.433 0.173 0.105 0.125 0.170 0.122 8636 2397 2400 2399 1440 10.548 0.471 --0.517 50.366 0.031 0.038 0.044 0.048 0.050 0.990 0.499 --0.500 8.482 0.174 0.191 0.205 0.213 0.217 4795 4795 --4795 4795 4795 4795 4795 4795 4795 10.729 0.325 0.289 0.142 0.532 56.811 0.025 0.031 0.047 0.052 0.057 1.002 0.469 0.453 0.349 0.499 11.287 0.157 0.174 0.211 0.222 0.232 8636 8636 8636 8636 8636 8636 8636 8636 8636 8636 8636 53.523 11.737 8.983 0.067 7171.387 540.048 0.001 97.006 0.174 21.279 8.523 11.901 0.038 1343.089 276.353 0.000 42.054 0.090 4795 4795 4795 4795 4795 4795 4795 4795 4795 53.778 11.562 8.945 0.066 7206.147 535.182 0.001 99.779 0.172 21.153 8.430 11.787 0.037 1353.573 272.569 0.000 41.706 0.090 8636 8636 8636 8636 8636 8636 8636 8636 8636 Individual Characteristics Education 1960 Dummy 1970 Dummy 1990 Dummy Female Age Born in 1905 Born in 1910 Born in 1915 Born in 1920 Born in 1925 State of Birth Characteristics % Urban % Foreign % Black % Emp.in Mfg. Ann. Mfg. Wage Val. of Farm per Acre P.C. # of Doctors P.C Educ. Expenditures # Schl Bldgs/ Sq. Mile Notes : Summary statistics are for state of birth, cohort and gender cells. All means and standard deviations use sample weights where the weights are the population estimates for the cell in the base period. Table 2: Summary Statistics for SIPP sample Variables Outcomes Self Reported Health Poor Health Fair or Poor Health Health Index Hospitalized in Last Year Days in Bed, last 4 months Number of Times Hospitalized Number of Nights in Hospital Trouble Seeing Trouble Hearing Trouble Speaking Trouble Lifting Trouble Walking Trouble with Stairs Trouble Getting Around Outside the Home Trouble Getting Around Inside the Home Trouble Getting In/Out of Bed Trouble Seeing at all Trouble Hearing at all Trouble Speaking at all Trouble Lifting at all Trouble Walking at all Trouble with Stairs at all Needs Help Getting Around Outside Needs Help Getting Around Inside Needs Help Getting In/Out of Bed Work limitation due to health conditions Arthritis Back Blind Cancer Deaf Deformity Diabetes Heart Hernia Hypertension Kidney Lung Mental Illness Missing Limb Paralysis Senility Stomach Stroke Thyroid Other Mean Std. dev. N 3.084 0.119 0.357 67.992 0.180 3.937 0.282 1.908 0.136 0.152 0.021 0.237 0.289 0.276 0.129 0.059 0.079 0.023 0.013 0.003 0.115 0.154 0.116 0.088 0.024 0.025 0.423 0.129 0.062 0.026 0.016 0.023 0.027 0.030 0.090 0.006 0.036 0.005 0.043 0.005 0.003 0.006 0.007 0.010 0.021 0.003 0.066 1.138 0.324 0.479 24.842 0.384 17.030 1.029 7.898 0.342 0.359 0.144 0.425 0.453 0.447 0.335 0.235 0.270 0.149 0.114 0.052 0.319 0.361 0.321 0.283 0.154 0.156 0.494 0.335 0.242 0.159 0.125 0.149 0.162 0.170 0.287 0.080 0.185 0.067 0.203 0.067 0.056 0.075 0.084 0.099 0.144 0.056 0.247 26030 26030 26030 26030 26484 25223 22229 26274 20853 20845 20834 20837 20799 20820 17401 17643 17636 20811 20819 15138 20789 20723 20775 13610 13893 13868 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 19073 Table 2: Summary Statistics for SIPP sample Variables Individual Characteristics Education Female Age Notes : Mean Std. dev. N 11.432 0.580 72.079 3.208 0.494 5.606 26030 4795 4795 Table 3: Replicating Lleras-Muney’s Estimates of Effects of Education on Mortality Dependent variable is ten year mortality rate Lleras-Muney (2005) Replication Replication with Wrong Base Year WLS IV WLS IV WLS IV Education -0.017 (0.004) -0.051 (0.026) -0.036 (0.004) -0.072 (0.025) -0.016 (0.004) -0.059 (0.026) Female -0.074 (0.003) -0.071 (0.004) -0.072 (0.003) -0.067 (0.004) -0.074 (0.003) -0.070 (0.004) % Urban -1.0E-04 -(9.4E-04) 0.001 (0.001) 9.3E-04 (9.9E-04) 0.002 (0.001) 4.4E-04 (1.0E-03) 0.002 (0.001) % Foreign -5.6E-04 (0.002) -0.0001 (0.002) -0.001 (0.002) -0.0005 (0.002) -0.002 (0.002) -0.0011 (0.002) % Black -0.002 (0.002) -0.0009 (0.002) -8.1E-04 (0.002) -5.9E-05 (0.002) -0.001 (0.002) -5.2E-04 (0.002) % employed in mfg. -0.071 (0.101) -0.11 (0.108) -9.3E-04 (0.236) -0.039 (0.246) 0.010 (0.234) -0.066 (0.246) Annual mfg. wage 7.4E-07 (3.1E-06) 0.000 (0.000) 3.4E-07 (4.1E-06) 4.7E-07 (4.3E-06) 3.0E-07 (4.0E-06) 5.6E-07 (4.3E-06) Val. of farm per acre 2.7E-06 (1.7E-05) 0.000 (0.000) -1.2E-06 (1.9E-05) -8.9E-06 (2.0E-05) 3.6E-06 (1.8E-05) -4.9E-06 (2.0E-05) Per capita # of doctors 0.242 (13.891) 7.926 (15.059) 16.394 (20.993) 42.372 (26.445) -0.511 (20.897) 26.405 (25.656) Per capita education exp. 1.9E-05 (7.9E-05) 0.000 (0.000) 5.3E-05 (8.3E-05) 6.2E-05 (9.4E-05) 4.4E-05 (7.6E-05) 4.5E-05 (8.5E-05) # school bldgs/sq. mile -0.008 (0.062) -0.005 (0.065) -0.0135001 (0.063) -0.012 (0.067) -0.015 (0.062) -0.013 (0.065) N R squared 4792 0.3575 4792 -- 4792 0.3606 4792 0.3536 4792 0.3549 4792 0.3425 Individual Characteristics State of Birth Characteristics Notes : All specifications include a dummy for 1970, 24 cohort dummies, 47 state of birth dummies, region of birth interacted with cohort and an intercept. Estimates are weighted using the number of observations in the cell in the base year. Standard errors, shown in parentheses, are clustered at the state of birth and cohort level. Table 4: New Estimates of Effects of Education on Mortality Dependent variable is ten year mortality rate, Table entries are the Coefficient on Education Goldin/Katz Sample and specification WLS IV N Instruments Panel A (1960-80) (1) 1% 1960-1980 -0.036 -0.072 4792 -(0.004) (0.025) (2) 1% 1960, 3% 1970, 5% 1980 drop allocated age, education, birthplace -0.045 (0.004) -0.043 (0.020) 4797 -0.045 (0.024) (3) Sample (2) with age cubic -0.039 (0.004) -0.046 (0.020) 4797 -0.047 (0.024) (4) Sample (2) with age cubic*yr -0.040 (0.004) -0.046 (0.020) 4797 -0.047 (0.024) (5) Sample (2) with state*cohort trend -0.048 (0.004) -0.032 (0.021) 4797 -0.016 (0.024) -0.034 (0.003) -0.035 (0.014) 8636 -0.026 (0.015) -0.036 (0.003) -0.020 (0.015) 8636 -0.012 (0.016) -0.025 (0.006) -0.085 (0.045) 2397 -0.081 (0.052) (9) Sample (6) 1970 only -0.061 (0.005) -0.022 (0.032) 2400 -0.023 (0.033) (10) Sample (6) 1980 only -0.043 (0.004) -0.006 (0.025) 2399 0.023 (0.029) (11) Sample (6) 1990 only -0.012 (0.005) 0.021 (0.040) 1440 0.027 (0.039) -0.017 (0.005) -0.059 (0.040) 2879 -0.067 (0.036) (13) Sample (6) 55-64 year olds -0.039 (0.005) -0.066 (0.041) 2398 0.063 (0.053) (14) Sample (6) 65-89 year olds -0.030 (0.003) -0.005 (0.019) 3071 -0.047 (0.023) -0.019 (0.004) -0.098 (0.037) 3644 -0.203 (0.125) -0.017 (0.004) 0.039 (0.022) 4992 0.025 (0.023) Panel B (1960-2000) (6) 1% 1960, 2% 1970, 5% 1980-00 age cubic*year (7) Sample (6) with state*cohort trend Panel C (1960-2000 by year) (8) Sample (6) 1960 only Panel D (1960-2000 by age) (12) Sample (6) 35-54 year olds Panel E (1960-2000 by cohort) (15) Sample (6) cohorts 1901-1912 (16) Sample (6) cohorts 1913-1925 Notes : All specifications include year dummies, cohort dummies, state of birth dummies, region of birth interacted with cohort and an intercept (except for rows 5 and 7). Estimates are weighted using the number of observations in the cell in the base year. Standard errors, shown in parentheses, are clustered at the state of birth and cohort level. Table 5: Estimates of Effects of Education on Health Outcomes in the SIPP IV /2SCML exogeneity test Dependent Variable OLS /Probit IV /2SCML % effect p-value N Panel A: General Health Outcomes Self Reported Health ( 1 is excellent, 5 is poor) -0.0941 (0.0023) -0.2289 (0.0745) -0.074 0.074 26030 Health Index (1 to 100 scale) 1.9674 (0.0511) 4.5345 (1.6738) 0.067 0.131 26030 Fair or Poor Health -0.0359 (0.0010) -0.0824 (0.0343) -0.230 0.176 26030 Poor Health -0.0141 (0.0006) -0.0269 (0.0206) -0.226 0.533 26030 Hospitalized in Last Year -0.0049 (0.0008) -0.0268 (0.0241) -0.149 0.364 26484 Days in Bed, last 4 months -0.3310 (0.0364) 2.1526 (1.4848) 0.547 0.074 25223 Number of Times Hospitalized -0.0101 (0.0024) -0.0944 (0.0884) -0.335 0.329 22229 Number of Nights in Hospital -0.0730 (0.0186) -1.0828 (0.7668) -0.567 0.185 26289 Panel B: Functional Limitations/ADL/IADL Trouble Seeing -0.0122 (0.0007) -0.0559 (0.0254) -0.412 0.085 20853 Trouble Hearing -0.0103 (0.0007) -0.0499 (0.0247) -0.329 0.109 20845 Trouble Speaking -0.0019 (0.0002) -0.0192 (0.0079) -0.909 0.039 20573 Trouble Lifting -0.0198 (0.0009) -0.0055 (0.0330) -0.023 0.667 20837 Trouble Walking -0.0251 (0.0011) 0.0130 (0.0325) 0.045 0.242 20797 Trouble with Stairs -0.0250 (0.0010) -0.0066 (0.0324) -0.024 0.993 20820 Trouble Getting Around Outside the Home -0.0120 (0.0008) -0.0146 (0.0257) -0.114 0.918 17401 Trouble Getting Around Inside the Home -0.0048 (0.0005) 0.0051 (0.0208) 0.087 0.635 17463 Trouble Getting In/ Out of Bed -0.0056 (0.0006) 0.0013 (0.0230) 0.016 0.764 17621 Trouble Seeing at all -0.0020 -0.0078 -0.343 0.490 20589 Table 5: Estimates of Effects of Education on Health Outcomes in the SIPP IV /2SCML exogeneity test OLS /Probit IV /2SCML (0.0002) (0.0084) Trouble Hearing at all -0.0008 (0.0001) -0.0100 (0.0045) Trouble Speaking at all 0.0000 (0.0001) -0.0008 ** Trouble Lifting at all -0.0100 (0.0007) -0.0029 (0.0250) Trouble Walking at all -0.0148 (0.0008) Trouble with Stairs at all Dependent Variable % effect p-value N -0.758 0.060 20256 0.000 7516 -0.025 0.775 20789 0.0107 (0.0260) 0.069 0.328 20723 -0.0114 (0.0006) 0.0071 (0.0202) 0.061 0.359 20775 Needs Help Getting Around Outside -0.0066 (0.0007) 0.0044 (0.0153) 0.050 0.470 13598 Needs Help Getting Around Inside -0.0010 (0.0002) 0.0108 (0.0078) 0.446 0.125 13757 Needs Help Getting In/Out of Bed -0.0011 (0.0003) 0.0092 (0.0080) 0.372 0.191 13794 Panel C: Specific Health Conditions Difficulty -0.0250 (0.0013) -0.0743 (0.0348) -0.175 0.157 19073 Arthritis -0.0088 (0.0008) -0.0043 (0.0217) -0.034 0.836 19012 Back -0.0028 (0.0005) -0.0349 (0.0167) -0.561 0.061 18924 Blind -0.0014 (0.0003) 0.0145 (0.0084) 0.557 0.060 18454 Cancer -0.0007 (0.0002) 0.0025 (0.0078) 0.161 0.677 18569 Deaf -0.0003 (0.0002) -0.0041 (0.0064) -0.179 0.568 18422 Deformity -0.0006 (0.0002) -0.0159 (0.0066) -0.591 0.018 18821 Diabetes -0.0023 (0.0003) -0.0258 (0.0082) -0.868 0.007 18688 Heart -0.0062 (0.0006) -0.0014 (0.0194) -0.016 0.804 19025 Hernia -0.0003 (0.0001) 0.0023 (0.0037) 0.362 0.454 17179 Table 5: Estimates of Effects of Education on Health Outcomes in the SIPP IV /2SCML exogeneity test OLS /Probit IV /2SCML -0.0031 (0.0004) 0.0376 (0.0124) Kidney -0.0001 (0.0001) Lung Dependent Variable Hypertension % effect 1.053 p-value 0.000 N 18683 0.0042 (0.0027) 0.938 0.072 16593 -0.0037 (0.0005) 0.0203 (0.0152) 0.472 0.106 19060 Mental Illness -0.00009 (0.00008) -0.0002 (0.0424) -0.045 0.932 15794 Missing Limb -0.00007 (0.00005) -0.0019 (0.0016) -0.580 0.155 14565 Paralysis -0.00011 (0.00006) 0.0016 (0.0020) 0.287 0.348 17301 Senility -0.00005 (0.00002) -0.0015 (0.0006) -0.214 0.070 17993 Stomach -0.0006 (0.0002) 0.0069 (0.0060) 0.695 0.195 17701 Stroke -0.0008 (0.0003) 0.0084 (0.0090) 0.397 0.295 18918 -0.0000001 (0.000000) 0.000001 ** 0.000 0.000 14559 -0.0023 (0.0005) -0.0013 (0.0152) -0.019 0.947 19060 Thyroid Other Table 6: Small Pox Vaccination Rates of Young Children in 1930 state Alabama Alaska Arizona. Arkansas. California. Colorado. Connecticut Delaware. DC Florida Georgia Hawaii Idaho Illinois. Indiana Iowa. Kansas Kentucky. Louisiana Maine. Maryland. Massachusetts Michigan. Minnesota Mississippi. Missouri Montana Nebraska Nevada. NH NJ NM NY NC ND Ohio. Oklahoma Oregon Pennsylvania RI SC SD Tennessee. Texas Utah. Vermont Virginia Washington WV Wisconsin Wyoming. Median at age 4 9 NA 22 5 23 13 35 8 14 NA 30.5 33 11 16 13 18 14 20.5 23 28 34 25 17 10 21.5 21 9 16 28 28 25 NA 23 3.5 33 15 19 13 9 51 11 25 10 13 13 NA 10 14.5 NA 18 NA 17.0 at age 5 15 NA 25 23 33 53 65 4 35 NA 56.5 38 18 26 14 26 26 50 46 50 60 62 25 15 31.5 37 8 15 28 76 53 NA 63 10 37 34 27 15 29 86 17 40 23 27 13 NA 16 24.5 NA 27 NA 27.0 change, age 4 to 5 % change, age 4 to 5 6 NA 3 18 10 40 30 -4 21 NA 26 5 7 10 1 8 12 29.5 23 22 26 37 8 5 10 16 -1 -1 0 48 28 NA 40 6.5 4 19 8 2 20 35 6 15 13 14 0 NA 6 10 NA 9 NA 10.0 66.7 NA 13.6 360.0 43.5 307.7 85.7 -50.0 150.0 NA 85.2 15.2 63.6 62.5 7.7 44.4 85.7 143.9 100.0 78.6 76.5 148.0 47.1 50.0 46.5 76.2 -11.1 -6.3 0.0 171.4 112.0 NA 173.9 185.7 12.1 126.7 42.1 15.4 222.2 68.6 54.5 60.0 130.0 107.7 0.0 NA 60.0 69.0 NA 50.0 NA 66.7 Table 7: IV Estimates of Mortality and Health by Stringency of Compulsory Vaccination Laws Mortality -0.028 (0.014) 8636 Health Index 4.535 (1.674) 26030 Poor Health -0.035 (0.024) 26030 Trouble Seeing -0.057 (0.028) 20853 Trouble Hearing -0.053 (0.027) 20845 Trouble Speaking -0.022 (0.012) 20834 Back -0.035 (0.019) 19073 Sample with Vacc data -0.025 (0.015) 7736 4.854 (1.590) 24958 -0.046 (0.023) 24958 -0.050 (0.027) 20045 -0.052 (0.026) 20036 -0.020 (0.011) 20027 -0.032 (0.019) 18371 -0.020 (0.012) 18371 -0.012 (0.007) 18371 Stringent States -0.023 (0.012) 3600 3.348 (1.804) 13841 -0.059 (0.027) 13841 -0.029 (0.030) 11225 -0.038 (0.024) 11219 -0.002 (0.007) 11207 -0.018 (0.020) 10417 -0.014 (0.012) 10417 -0.001 (0.006) 10417 Non-Stringent States 0.020 (0.027) 4136 3.774 (1.760) 11117 0.001 (0.029) 11117 -0.023 (0.033) 8820 0.026 (0.032) 8817 -0.009 (0.012) 8820 -0.030 (0.029) 7954 -0.016 (0.018) 7954 -0.022 (0.014) 7954 Baseline Sample Stiffness or deformity -0.021 (0.012) 19073 Senility -0.013 (0.007) 19073 Notes: All IV regressions include female dummy, cohort dummies, state of birth dummies, and 7 time varying state of birth characteristics (at age 14) from Lleras-Muney (2005). These are % urban, % foreign born, %black, %mfg, mfg wage, doctors per-capita, education expenditures per-capita, and schools per sq. mile. The mortality results also use region of birth interacted with cohort while the SIPP results use state-specific cohort trends. Instruments are categories of required years of schooling in state of residence at age 14. Standard errors are clustered on state of birth and cohort Figure 1: Ten Year Mortality Rates by Age Across Census Years 0.8 0.7 0.6 . ·. 0.4 I~ 0.3 0.2 0.1 Age 79 77 75 73 71 69 67 65 63 61 59 57 55 53 51 49 47 45 43 41 39 37 0 35 Death Rate 0.5 1960 1970 1980 1990 I Figure 2: Vaccination Rates for 4 and 5 year olds by state 70 age 4 age 5 60 50 40 30 20 10 0 Massachusetts Minnesota Utah. Kentucky. Median State Working Paper Series A series of research studies on regional economic issues relating to the Seventh Federal Reserve District, and on financial and economic topics. A Proposal for Efficiently Resolving Out-of-the-Money Swap Positions at Large Insolvent Banks George G. Kaufman WP-03-01 Depositor Liquidity and Loss-Sharing in Bank Failure Resolutions George G. Kaufman WP-03-02 Subordinated Debt and Prompt Corrective Regulatory Action Douglas D. Evanoff and Larry D. Wall WP-03-03 When is Inter-Transaction Time Informative? Craig Furfine WP-03-04 Tenure Choice with Location Selection: The Case of Hispanic Neighborhoods in Chicago Maude Toussaint-Comeau and Sherrie L.W. Rhine WP-03-05 Distinguishing Limited Commitment from Moral Hazard in Models of Growth with Inequality* Anna L. Paulson and Robert Townsend WP-03-06 Resolving Large Complex Financial Organizations Robert R. Bliss WP-03-07 The Case of the Missing Productivity Growth: Or, Does information technology explain why productivity accelerated in the United States but not the United Kingdom? Susanto Basu, John G. Fernald, Nicholas Oulton and Sylaja Srinivasan WP-03-08 Inside-Outside Money Competition Ramon Marimon, Juan Pablo Nicolini and Pedro Teles WP-03-09 The Importance of Check-Cashing Businesses to the Unbanked: Racial/Ethnic Differences William H. Greene, Sherrie L.W. Rhine and Maude Toussaint-Comeau WP-03-10 A Firm’s First Year Jaap H. Abbring and Jeffrey R. Campbell WP-03-11 Market Size Matters Jeffrey R. Campbell and Hugo A. Hopenhayn WP-03-12 The Cost of Business Cycles under Endogenous Growth Gadi Barlevy WP-03-13 The Past, Present, and Probable Future for Community Banks Robert DeYoung, William C. Hunter and Gregory F. Udell WP-03-14 1 Working Paper Series (continued) Measuring Productivity Growth in Asia: Do Market Imperfections Matter? John Fernald and Brent Neiman WP-03-15 Revised Estimates of Intergenerational Income Mobility in the United States Bhashkar Mazumder WP-03-16 Product Market Evidence on the Employment Effects of the Minimum Wage Daniel Aaronson and Eric French WP-03-17 Estimating Models of On-the-Job Search using Record Statistics Gadi Barlevy WP-03-18 Banking Market Conditions and Deposit Interest Rates Richard J. Rosen WP-03-19 Creating a National State Rainy Day Fund: A Modest Proposal to Improve Future State Fiscal Performance Richard Mattoon WP-03-20 Managerial Incentive and Financial Contagion Sujit Chakravorti and Subir Lall WP-03-21 Women and the Phillips Curve: Do Women’s and Men’s Labor Market Outcomes Differentially Affect Real Wage Growth and Inflation? Katharine Anderson, Lisa Barrow and Kristin F. Butcher WP-03-22 Evaluating the Calvo Model of Sticky Prices Martin Eichenbaum and Jonas D.M. Fisher WP-03-23 The Growing Importance of Family and Community: An Analysis of Changes in the Sibling Correlation in Earnings Bhashkar Mazumder and David I. Levine WP-03-24 Should We Teach Old Dogs New Tricks? The Impact of Community College Retraining on Older Displaced Workers Louis Jacobson, Robert J. LaLonde and Daniel Sullivan WP-03-25 Trade Deflection and Trade Depression Chad P. Brown and Meredith A. Crowley WP-03-26 China and Emerging Asia: Comrades or Competitors? Alan G. Ahearne, John G. Fernald, Prakash Loungani and John W. Schindler WP-03-27 International Business Cycles Under Fixed and Flexible Exchange Rate Regimes Michael A. Kouparitsas WP-03-28 Firing Costs and Business Cycle Fluctuations Marcelo Veracierto WP-03-29 Spatial Organization of Firms Yukako Ono WP-03-30 Government Equity and Money: John Law’s System in 1720 France François R. Velde WP-03-31 2 Working Paper Series (continued) Deregulation and the Relationship Between Bank CEO Compensation and Risk-Taking Elijah Brewer III, William Curt Hunter and William E. Jackson III WP-03-32 Compatibility and Pricing with Indirect Network Effects: Evidence from ATMs Christopher R. Knittel and Victor Stango WP-03-33 Self-Employment as an Alternative to Unemployment Ellen R. Rissman WP-03-34 Where the Headquarters are – Evidence from Large Public Companies 1990-2000 Tyler Diacon and Thomas H. Klier WP-03-35 Standing Facilities and Interbank Borrowing: Evidence from the Federal Reserve’s New Discount Window Craig Furfine WP-04-01 Netting, Financial Contracts, and Banks: The Economic Implications William J. Bergman, Robert R. Bliss, Christian A. Johnson and George G. Kaufman WP-04-02 Real Effects of Bank Competition Nicola Cetorelli WP-04-03 Finance as a Barrier To Entry: Bank Competition and Industry Structure in Local U.S. Markets? Nicola Cetorelli and Philip E. Strahan WP-04-04 The Dynamics of Work and Debt Jeffrey R. Campbell and Zvi Hercowitz WP-04-05 Fiscal Policy in the Aftermath of 9/11 Jonas Fisher and Martin Eichenbaum WP-04-06 Merger Momentum and Investor Sentiment: The Stock Market Reaction To Merger Announcements Richard J. Rosen WP-04-07 Earnings Inequality and the Business Cycle Gadi Barlevy and Daniel Tsiddon WP-04-08 Platform Competition in Two-Sided Markets: The Case of Payment Networks Sujit Chakravorti and Roberto Roson WP-04-09 Nominal Debt as a Burden on Monetary Policy Javier Díaz-Giménez, Giorgia Giovannetti, Ramon Marimon, and Pedro Teles WP-04-10 On the Timing of Innovation in Stochastic Schumpeterian Growth Models Gadi Barlevy WP-04-11 Policy Externalities: How US Antidumping Affects Japanese Exports to the EU Chad P. Bown and Meredith A. Crowley WP-04-12 Sibling Similarities, Differences and Economic Inequality Bhashkar Mazumder WP-04-13 3 Working Paper Series (continued) Determinants of Business Cycle Comovement: A Robust Analysis Marianne Baxter and Michael A. Kouparitsas WP-04-14 The Occupational Assimilation of Hispanics in the U.S.: Evidence from Panel Data Maude Toussaint-Comeau WP-04-15 Reading, Writing, and Raisinets1: Are School Finances Contributing to Children’s Obesity? Patricia M. Anderson and Kristin F. Butcher WP-04-16 Learning by Observing: Information Spillovers in the Execution and Valuation of Commercial Bank M&As Gayle DeLong and Robert DeYoung WP-04-17 Prospects for Immigrant-Native Wealth Assimilation: Evidence from Financial Market Participation Una Okonkwo Osili and Anna Paulson WP-04-18 Individuals and Institutions: Evidence from International Migrants in the U.S. Una Okonkwo Osili and Anna Paulson WP-04-19 Are Technology Improvements Contractionary? Susanto Basu, John Fernald and Miles Kimball WP-04-20 The Minimum Wage, Restaurant Prices and Labor Market Structure Daniel Aaronson, Eric French and James MacDonald WP-04-21 Betcha can’t acquire just one: merger programs and compensation Richard J. Rosen WP-04-22 Not Working: Demographic Changes, Policy Changes, and the Distribution of Weeks (Not) Worked Lisa Barrow and Kristin F. Butcher WP-04-23 The Role of Collateralized Household Debt in Macroeconomic Stabilization Jeffrey R. Campbell and Zvi Hercowitz WP-04-24 Advertising and Pricing at Multiple-Output Firms: Evidence from U.S. Thrift Institutions Robert DeYoung and Evren Örs WP-04-25 Monetary Policy with State Contingent Interest Rates Bernardino Adão, Isabel Correia and Pedro Teles WP-04-26 Comparing location decisions of domestic and foreign auto supplier plants Thomas Klier, Paul Ma and Daniel P. McMillen WP-04-27 China’s export growth and US trade policy Chad P. Bown and Meredith A. Crowley WP-04-28 Where do manufacturing firms locate their Headquarters? J. Vernon Henderson and Yukako Ono WP-04-29 Monetary Policy with Single Instrument Feedback Rules Bernardino Adão, Isabel Correia and Pedro Teles WP-04-30 4 Working Paper Series (continued) Firm-Specific Capital, Nominal Rigidities and the Business Cycle David Altig, Lawrence J. Christiano, Martin Eichenbaum and Jesper Linde WP-05-01 Do Returns to Schooling Differ by Race and Ethnicity? Lisa Barrow and Cecilia Elena Rouse WP-05-02 Derivatives and Systemic Risk: Netting, Collateral, and Closeout Robert R. Bliss and George G. Kaufman WP-05-03 Risk Overhang and Loan Portfolio Decisions Robert DeYoung, Anne Gron and Andrew Winton WP-05-04 Characterizations in a random record model with a non-identically distributed initial record Gadi Barlevy and H. N. Nagaraja WP-05-05 Price discovery in a market under stress: the U.S. Treasury market in fall 1998 Craig H. Furfine and Eli M. Remolona WP-05-06 Politics and Efficiency of Separating Capital and Ordinary Government Budgets Marco Bassetto with Thomas J. Sargent WP-05-07 Rigid Prices: Evidence from U.S. Scanner Data Jeffrey R. Campbell and Benjamin Eden WP-05-08 Entrepreneurship, Frictions, and Wealth Marco Cagetti and Mariacristina De Nardi WP-05-09 Wealth inequality: data and models Marco Cagetti and Mariacristina De Nardi WP-05-10 What Determines Bilateral Trade Flows? Marianne Baxter and Michael A. Kouparitsas WP-05-11 Intergenerational Economic Mobility in the U.S., 1940 to 2000 Daniel Aaronson and Bhashkar Mazumder WP-05-12 Differential Mortality, Uncertain Medical Expenses, and the Saving of Elderly Singles Mariacristina De Nardi, Eric French, and John Bailey Jones WP-05-13 Fixed Term Employment Contracts in an Equilibrium Search Model Fernando Alvarez and Marcelo Veracierto WP-05-14 Causality, Causality, Causality: The View of Education Inputs and Outputs from Economics Lisa Barrow and Cecilia Elena Rouse WP-05-15 5 Working Paper Series (continued) Competition in Large Markets Jeffrey R. Campbell WP-05-16 Why Do Firms Go Public? Evidence from the Banking Industry Richard J. Rosen, Scott B. Smart and Chad J. Zutter WP-05-17 Clustering of Auto Supplier Plants in the U.S.: GMM Spatial Logit for Large Samples Thomas Klier and Daniel P. McMillen WP-05-18 Why are Immigrants’ Incarceration Rates So Low? Evidence on Selective Immigration, Deterrence, and Deportation Kristin F. Butcher and Anne Morrison Piehl WP-05-19 Constructing the Chicago Fed Income Based Economic Index – Consumer Price Index: Inflation Experiences by Demographic Group: 1983-2005 Leslie McGranahan and Anna Paulson WP-05-20 Universal Access, Cost Recovery, and Payment Services Sujit Chakravorti, Jeffery W. Gunther, and Robert R. Moore WP-05-21 Supplier Switching and Outsourcing Yukako Ono and Victor Stango WP-05-22 Do Enclaves Matter in Immigrants’ Self-Employment Decision? Maude Toussaint-Comeau WP-05-23 The Changing Pattern of Wage Growth for Low Skilled Workers Eric French, Bhashkar Mazumder and Christopher Taber WP-05-24 U.S. Corporate and Bank Insolvency Regimes: An Economic Comparison and Evaluation Robert R. Bliss and George G. Kaufman WP-06-01 Redistribution, Taxes, and the Median Voter Marco Bassetto and Jess Benhabib WP-06-02 Identification of Search Models with Initial Condition Problems Gadi Barlevy and H. N. Nagaraja WP-06-03 Tax Riots Marco Bassetto and Christopher Phelan WP-06-04 The Tradeoff between Mortgage Prepayments and Tax-Deferred Retirement Savings Gene Amromin, Jennifer Huang,and Clemens Sialm WP-06-05 Why are safeguards needed in a trade agreement? Meredith A. Crowley WP-06-06 6 Working Paper Series (continued) Taxation, Entrepreneurship, and Wealth Marco Cagetti and Mariacristina De Nardi WP-06-07 A New Social Compact: How University Engagement Can Fuel Innovation Laura Melle, Larry Isaak, and Richard Mattoon WP-06-08 Mergers and Risk Craig H. Furfine and Richard J. Rosen WP-06-09 Two Flaws in Business Cycle Accounting Lawrence J. Christiano and Joshua M. Davis WP-06-10 Do Consumers Choose the Right Credit Contracts? Sumit Agarwal, Souphala Chomsisengphet, Chunlin Liu, and Nicholas S. Souleles WP-06-11 Chronicles of a Deflation Unforetold François R. Velde WP-06-12 Female Offenders Use of Social Welfare Programs Before and After Jail and Prison: Does Prison Cause Welfare Dependency? Kristin F. Butcher and Robert J. LaLonde Eat or Be Eaten: A Theory of Mergers and Firm Size Gary Gorton, Matthias Kahl, and Richard Rosen Do Bonds Span Volatility Risk in the U.S. Treasury Market? A Specification Test for Affine Term Structure Models Torben G. Andersen and Luca Benzoni WP-06-13 WP-06-14 WP-06-15 Transforming Payment Choices by Doubling Fees on the Illinois Tollway Gene Amromin, Carrie Jankowski, and Richard D. Porter WP-06-16 How Did the 2003 Dividend Tax Cut Affect Stock Prices? Gene Amromin, Paul Harrison, and Steven Sharpe WP-06-17 Will Writing and Bequest Motives: Early 20th Century Irish Evidence Leslie McGranahan WP-06-18 How Professional Forecasters View Shocks to GDP Spencer D. Krane WP-06-19 Evolving Agglomeration in the U.S. auto supplier industry Thomas Klier and Daniel P. McMillen WP-06-20 Mortality, Mass-Layoffs, and Career Outcomes: An Analysis using Administrative Data Daniel Sullivan and Till von Wachter WP-06-21 7 Working Paper Series (continued) The Agreement on Subsidies and Countervailing Measures: Tying One’s Hand through the WTO. Meredith A. Crowley How Did Schooling Laws Improve Long-Term Health and Lower Mortality? Bhashkar Mazumder WP-06-22 WP-06-23 8