Attending college has been considered one of the most pro table investment decisions, asits estimated annualized return ranges from 8% to 13%. However, a large fraction of high schoolgraduates do not enroll in college. Using a simple risk premium approach, I reconcile the observedhigh average returns to schooling with relatively low attendance rates. A high dropout risk hastwo important e ects on the estimated average returns to college: selection bias and risk premium.Once taking into account dropout risk, a simple calculation of risk premium accounts for 51% ofthe excess of return to college education. In order to explicitly consider the selection bias, I furtherexplore the dropout risk in a life-cycle model with heterogeneous ability. The risk-premium ofcollege participation accounts for 29% of the excess of returns to college education for high-abilitystudents, and accounts for 27% of the excess return for low-ability students, since they face alarger college dropout risk. Risk averse agents are willing to reduce their return to college inorder to avoid the dropout risk. The e ect is not uniform across ability levels.

This project laid the groundwork for an Internet-delivered Public ParticipationGeographic Information System to facilitate exploration and discovery of the pastcommunities of the Mammoth Cave Park area. The emergence of Internet Web 2.0design along with distributed GIS services allows for anyone to interact with and add tothe information found on central Internet sites. Historical geography often relies uponpublic participation from individuals outside the academic world to provide narrativedescriptions, photographs and manuscripts of past places and events to augmentinformation held by institutions and academia. A public-participation website for theMammoth Cave Historic GIS (MCHGIS) created a central Internet location for dispersedand disparate data related to pre-park communities to be presented with a geographiccontext. The MCHGIS project allowed for visualization of the pre-park communities inunique ways and contributed new understandings of this pre-park area.

Existing studies investigating the effect of mortality decline on retirement age usuallyspecify the human mortality pattern in a parametric way and consider the derivative ofoptimal retirement age with respect to a change in the survival parameter. However, asurvival parameter change affects the mortality rates at different ages simultaneously.Motivated by the stylized fact that mortality decline affects primarily younger people inthe early phase of a mortality transition but mainly older people in the later phase, westudy the more fundamental question of how a mortality change at an arbitrary age affectsoptimal retirement age. By using the Volterra derivative for a functional, we show that amortality decline at an older age unambiguously leads to a later retirement age, but amortality decline at a younger age may lead to earlier retirement because of thesubstantial increase in the individual's expected lifetime human wealth.

We propose a theory of skill mobility across cities. It predicts the well documented city sizewage premium: the wage distribution in larger and more productive cities first-order stochastically dominates that in less productive cities. Yet, because this premium reflects higher house prices, this does not necessarily imply that this stochastic dominance relation also exists in the distribution of skills. Our model predicts quite to the contrary: instead of first-order, there is second-order stochastic dominance in the skill distribution. The demand for skills is non-monotonic as our model predicts a Sinatra as well as an Eminem effect: both the very high and the very low skilled disproportionately sort into the biggest cities, while those with medium skill levels sort into small cities. Based on our theory, the pattern of spatial sorting is explained by a simple technology with varying elasticity of substitution by skill. Using CPS data on wages and Census data on house prices, this technology with the elasticity of substitution decreasing in skill density is consistent with the observed patterns of skills.

This paper introduces a new dataset detailing the task composition of occupations inthe United States labor-force for the period 1880-1940, using information about the taskcontent of over 4,000 occupations from the Dictionary of Occupational Titles (1939 and1949). This unique data is used to analyze the eect of electri?cation on the task content ofUS manufacturing. Preliminary estimates show that electri?cation increased the demandfor clerical, numerical, planning and people skills relative to that for manual skills andthe ability to manipulate machinery. Thus early twentieth century technological changewas unskill-biased at the factory ?oor level but skill-biased on aggregate. The ?ndingssuggest that the "hollowing out" in the skill distribution that has been found for the recentcomputerization era is not a new phenomenon.

Creation, (re)use and discovery of research data often follows a predictable flow, suggesting a lifecycle of research data. Locating needed social science data in most currently available archive catalogs is suboptimal as they do not provide searching functions that align with how researchers seek specific data. Exposing and indexing the holdings of data archives and publications in standardized metadata formats could enable web-scale discovery through new cross-collection search engine functions built to exploit that metadata.

This paper investigates recent commuting trends in American workers. Unlike most studies of commuting that rely on Census data, this study utilizes the unique American Time Use Survey to detail the complex commuting patterns of modern-day workers. The data confirm what has been suspected, that incidence of driving alone has decreased substantially in recent years while carpooling has rebounded. The results from the multi-nominal logistic estimation of workers’ commuting choices yield support for both the traditional economic determinants as well as for the newer, socio-economic factors. In addition to the cost savings, many commuters appear to value the social aspect of carpooling. Surprisingly, there is little evidence that the need for autonomy plays much of a factor in explaining worker’s choice of the journey to work. The estimated short-run “elasticity” of carpooling with respect to real gas prices appears to be quite high and largely accounts for the significant decline in the incidence of “driving alone”.

To what extent do politicians reward voters who are members of their own ethnic or racial group? Using data from large cities in the United States, we study how black employment outcomes are affected by changes in the race of the cities mayors between 1971 and 2003. We find that black employment and labor force participation rise, and the black unemployment rate falls, during the tenure of black mayors both in absolute terms and relative to whites. Black employment gains in municipal government jobs are particularly large, which suggests that our results capture the causal effects of black mayors. We also find that the effect of black mayors on black employment outcomes is stronger in cities that have a large black community. This suggests that electoral incentives may be an important determinant of racial favoritism. Finally, we also find that, corresponding to increases in employment, black income is higher after black mayors take office. Again, this effect is pronounced in cities with a large black population.

This study investigated the effects of widespread forest clearance and fragmentation on forest compositional change between pre-settlement and the present (ca. 18001993) in western New York. Forest compositional turnover metrics were calculated to evaluate whether soil conditions accounted for the forest compositional change, to investigate how forest clearance might have contributed to the change, and to compare with the amount of change noted in other studies that compared the pre-settlement land surveys and the current Forest Inventory and Analysis (FIA) surveys. The results suggested weak relations between the forest compositional change and soil conditions. The amount of change at the county level was positively correlated with the amount of forest clearance by the late 1800s and negatively correlated with the amount of remnant forest in the late 1800s. Furthermore, while the geographic distributions of late-successional taxa in the study area exhibited increased patchiness and decreased contagion, those of early-successional taxa exhibited decreased patchiness and increased contagion. Comparison of taxas changed abundances with seven life history characteristics commonly associated with early- and late successional status showed that taxas changed abundances were significantly related to maximum life and shade tolerance. Analysis of different-sized trees in different ages in the FIA survey indicated the possibility of the current forests trending back toward their pre-settlement composition.

Are households more likely to be homeowners when housing risk is higher? Weshow that homeownership rates and loan-to-value (LTV) ratios at the city level arestrongly negatively correlated with house price levels and the variance of house pricegrowth rates in the city. But both price levels and the variance of their growth rates arethemselves correlated with the relative value of land in the city, even when land valueis instrumented for using topographic measures. We disentangle the contributions ofhigh prices from high variances by building a life-cycle model of homeownership choices.The model is able to explain much of the cross-city dispersion in homeownership andLTV. We find that higher price levels explain the lower homeownership while higherrisk explains the lower LTV in high land value cities. The variation in LTV withrisk highlights the importance of including other means of insurance in models ofhomeownership.

The Great Recession is affecting everyone in one wayor another, but not everyone is affected in the sameway. Womens and mens work (both in and out of thelabor force) still differs, so we can expect that the economiccrisis has had a distinct impact on women aswell as their families. This policy brief discusses howthe down economy has differentially impacted womenand men in Massachusetts and the gendered implicationsof federal stimulus spending. It also identifiespotential opportunities to promote gender equality asthe United States, and Massachusetts in particular,attempt to move beyond the Great Recession.

Borrowers in cities where house prices boomed in the 2000s relied heavily on backloaded interest-only (IO) mortgages that require borrowers only to pay interest initially. We develop a theory that encompasses common explanations for IO use and show that while they can largely account for the regional variation in IOs, they cannot fully explain the concentration of IOs in booming cities. We propose a new explanation. In our model, uncertain price appreciation and no-recourse lending can lead to speculation financed with backloaded mortgages. We find evidence that IO borrowers behaved in ways consistent with such speculation.

Through a comparative historical analysis of the American states, I show how public education was the original policy field through which white American women became empowered as voters and political officials. Women's changing status within the education profession and school suffrage rights are an important and overlooked aspect of women's political history, and the rural orientation of state governments and women's increasing administrative authority as county superintendents and rural supervisors of education was pivotal to women's political empowerment. Women's authority, however, varied across regions and across states, with women's authority especially strong in Western states. I find that women in the field of public education were most empowered where there was a history of school suffrage rights, where administrative offices were elective rather than appointed, and where the power of the state superintendent of public instruction was weak. These findings suggest that democratic institutions, more than economic development or state capacity, were fundamental to women's increasing authority in the policy domain that commanded the largest share of state and local resources at the time.

Many statistical agencies disseminate samples of census microdata, that is, data on individual records, to the public. Before releasing the microdata, agencies typically alter identifying or sensitive values to protect data subjects confidentiality, for example by coarsening, perturbing, or swapping data. These standard disclosure limitation techniques distort relationships and distributional features in the original data, especially when applied with high intensity. Furthermore, it can be difficult for analysts of the masked public use data to adjust inferences for the effects of the disclosure limitation. Motivated by these shortcomings, we propose an approach to census microdata dissemination called sampling with synthesis. The basic idea is to replace the identifying or sensitive values in the census with multiple imputations, and release samples from these multiply-imputed populations. We demonstrate that sampling with synthesis can improve the quality of public use data relative to sampling followed by standard statistical disclosure limitation; simulation results showing this are available online as supplemental material. We derive methods for analyzing the multiple datasets generated by sampling with synthesis. We present algorithms for selecting which census values to synthesize based on considerations of disclosure risk and data utility. We illustrate sampling with synthesis on a population constructed with data from the U.S. Current Population Survey.