We present simple procedures for estimating non linear panel data models in the presence of unobserved heterogeneity and possible endogeneity with respect to time-varying unobservables. We combine a correlated random effects approach with a control function approach while accounting for missing time periods for some units. We examine the performance of the approach in comparisons with standard estimators using Monte Carlo simulation. We apply the methods to estimate the effects of school spending on student pass rates on a standardized math exam. We find that a 10% increase in spending leads to an approximately 2 percentage point increase in math pass rates.

I analyze the effects of public pension parameters, Social Security coverage, and state pension fund sustainability on the retirement of public employees. I use data from the Health and Retirement Study, including personal early and normal retirement eligibility and state of residence. I develop a state-level measure of effective public pension plan sustainability that reflects both the degree of public plan underfunding and a state’s ability to fund the plan with its own resources. Using the Public Plans Database and the Treasury Department’s estimate of Total Taxable Resources, I calculate the state tax rate that, applied to a state’s total taxable resources, could fund the state’s unfunded actuarial accrued liability. This effective tax rate varies by Social Security status of the plan. I model retirement probability as a function of public pension eligibility, Social Security coverage in the public sector job, and effective underfunding. I find that becoming eligible for early or normal retirement, or receiving an early-out offer, significantly increases the probability of retiring beginning at age 50. Having Social Security coverage approximately doubles this probability. Public sector workers without Social Security coverage are estimated to have a lower probability of retirement at key eligibility ages. I find that the probability of retirement falls with the degree of underfunding or effective plan risk, but this effect is small compared to the response to plan features. These findings suggest that state legislative action to affect retirement decisions would be most effective operating through plan eligibility rules.

For the panel data case where cross-sectional units are nested within higher-level groups, and there are many such groups, we propose a test that allows one to determine whether controlling for fixed effects at the more aggregate level is sufficient. The alternative is that one should allow for fixed effects at the unit level. The regression-based test is simple to carry out, even for unbalanced panels. In addition, the test is easily made robust to arbitrary heteroskedasticity, serial correlation across time, and even cluster correlation at the group level. We also show how to modify the traditional Hausman test of a single coefficient to be fully robust to serial correlation and cluster correlation. The tests work well in terms of size and power in a small simulation study. We apply the test to choosing between a fixed effects analysis at the school district level and the disaggregated school level.

AbstractI analyze the effects of state public pension parameters on the retirement of public employees. Using a panel data set of public sector workers from 12 waves of the Health and Retirement Study, I model the probability of retirement as a function of pension wealth at early and normal retirement eligibility and Social Security coverage in the public sector job. I find that becoming eligible for early retirement, or receiving an early-out offer, significantly increases the probability of retiring. I do not find any effect of retirement wealth levels. These findings suggest that state legislative action to affect retirement decisions and reduce future pension costs would be most effective operating through plan eligibility rules and early-out incentives.

An abstract is not available for this content so a preview has been provided. Please use the Get access link above for information on how to access this content.

Enterprise zones are geographically targeted economic development incentives used in the United States by individual states since the early 1980s and the federal government since 1993. Research on state zone programmes that accounts for the endogeneity of zone designation finds little improvement in the employment and incomes of zone residents, but some evidence that firms respond to tax incentives for capital. In contrast, the federal empowerment zone programme combines tax incentives with local initiatives and access to large federal grants. Recent research on round one of the federal programme finds mixed evidence on zone resident employment.

I use district-level panel data from the 1991/92 school year through the 2003/04 school year to study the effects of Proposal A, passed in Michigan in 1994. Proposal A dramatically changed the way schools were funded starting in the 1994/95 school year. I discuss what has happened to the pattern of spending in years before and after the reform, including a discussion of funding equalization. Using the several additional years of data—which include additional periods of substantial funding increases for low-spending districts—and a richer lag structure in an econometric model, I find that increases in spending have nontrivial, statistically significant effects on math test pass rates. The effects are notably larger for districts with initially poor performance.

We revisit the effects of spending on student performance using data from the state of Michigan. In addition to exploiting a dramatic change in funding in the mid-1990s and subsequent nonsmooth changes, we propose nonlinear panel data models that recognize the bounded nature of the pass rate. Importantly, we show how to estimate average partial effects, which can be compared across many different models (linear and nonlinear) under different assumptions and estimated using many different methods. We find that spending has nontrivial and statistically significant effects, although the diminishing effect is not especially pronounced.

We explore the interaction of state pension systems with state finances. We find that changes in pension assets are an important source of funding for state governments, but that states face incentive problems that impede funding efforts with the result that many plans are underfunded. We analyze the substantive differences between defined benefit and defined contribution plans for public employees and state governments, Regression analysis using a panel of 85 state public pension plans indicates some evidence of actuarial assumption manipulation to reduce funding pressure. Plan demographics and state tax revenues are significant influences on funding ratios, while plan features are not.

We show how to compute the standard error for a nonlinear function of regression coefficients using a simple substitution trick. We use the method to obtain a standard error for the long-run effect in a dynamic panel data model.

This study uses data on standardized test scores from 1992 through 1998 at Michigan schools to determine the effects of spending on student performance. The years in the data set straddle 1994, when Michigan dramatically changed the way that K-12 schools are funded, and moved toward equalization of spending across schools. Focusing on pass rates for a fourth-grade math tests (the most complete and consistent data available for Michigan), I find that increases in spending have nontrivial, statistically significant effects on math test pass rates, and the effects are largest for schools with initially poor performance.

In 1997, the State of Michigan closed its DB pension plan to new state employees. New employees are automatically enrolled in a 401(k) plan with a mandatory state contribution. Existing employees chose between staying in the DB plan, or transferring the present value of their vested pension benefits to a DC plan. This paper surveys the event and analyzes the choice made by corrections workers, about 25 percent of Michigan public employment. An analysis of the switch to individual accounts may inform Federal estimates of take-up rates in a possible privatization of Social Security.

Understanding the role that 401(k) plan characteristics – like investment choice – play in participation and employee contributions is important as more workers rely on this type of retirement plan and proposals for Social Security solvency include individual savings plans. Using the 1992 Health and Retirement Study, this paper investigates which individual and job characteristics are associated with asset choice in defined contribution plans. Investment choice is found to substantially increase contributions to defined contribution plans.

Workers with individual retirement saving accounts often make decisions about contribution rates and asset allocation that affect ultimate retirement income. This paper presents econometric evidence from two data sets on the role that participant investment choice plays in asset allocation, contributions, and account balances. My preferred estimates indicate that participants with investment choice are 36 percent more likely to make an annual contribution. These participants are estimated to invest 13 percentage points more in stocks, contribute between one and three percentage points more of salary, and have at least $9000 more in their account than comparable participants without investment choice.

This paper chronicles the experiences of the U.S. withholding tax on interest income. In 1984, the U.S. repealed its 30 percent withholding tax on interest income paid to foreign persons or corporations. While the tax raised little revenue, it had imposed substantial implicit costs on U.S. corporate borrowers. Since, prior to repeal, domestically issued bonds were subject either to withholding or strict information requirements, many U.S. multinationals raised funds through foreign finance subsidiaries, primarily in the Netherlands Antilles, to avoid the tax. Although the withholding tax rate was effectively reduced to zero in the U.S., this paper demonstrates that interest flows were highly sensitive to their after-tax cost.