In this paper we forecast social security disability applications in the USA using a spatial dynamic panel data model. Specifically, we forecast three types of applications—those who filed under (i) the Social Security Disability Insurance (SSDI), (ii) the Supplemental Security Income (SSI), and (iii) simultaneously (Concurrent). We find strong cross-sectional correlations in claim applications across states, which is exploited in forecasting. Depending on the forecast horizons from 1 to 12 months, the improvement in out-of-sample state level forecast accuracies as measured by root mean squared error (RMSE) is substantial, and find that the contemporaneous spatial lag term has the greatest contribution to the improvement. The gain from the spatial lag increases with forecast horizon—at the 9-month horizon the aggregate RMSE is reduced by almost 28

We unravel two drastically different paths for the uninsured and for people with employer-sponsored health insurance (ESI) to cope with the economic consequences of health shocks. The uninsured individuals support their incomes mostly from other non-labor income sources, public transfers, and borrowing. The overall budget constraint of the uninsured tightens to the extent that they are not able to adequately increase medical consumption and maintain necessary consumption in response to shocks. In contrast, people with ESI maintain most of their income by keeping their jobs, increase their medical expenses without sacrificing necessary consumption and are more likely to recover from illness. Overall, our results highlight the economic and health benefits of health insurance in response to health shocks.

This paper examines the seasonality in the U.S. Social Security disability applications, and shows that the monthly disability applications exhibit a double-peak seasonal pattern which lags a similar seasonal pattern in unemployment and unemployment insurance initial claims by one to two months. The broad seasonal patterns in disability applications are remarkably similar across states but with significant heterogeneity in amplitudes, which seems to be associated with climatic factors. We utilize this inter-state heterogeneity to show that the seasonal patterns in disability applications and labor market conditions are correlated even after controlling for climatic effects. We also show that the seasonally in disability applications generated by the automatic settings of the widely used seasonal adjustment program X13 ARIMA-SEATS are distorted during the COVID-19 pandemic and the pattern of the distortion is similar to that in employment data.

Judging the conformity of binary events in macroeconomics and finance has often been done with indices that measure synchronization. In recent years, the use of Receiver Operating Characteristic (ROC) curve has become popular for this task. This article shows that the ROC and synchronization approaches are closely related, and each can be derived from a decision-making framework. Furthermore, the resulting global measures of the degree of conformity can be identified and estimated using the standard method of moments estimators. The impact of serial dependence in the underlying series upon inferences can therefore be allowed for. Such serial correlation is common in macroeconomic and financial data.

This paper considers bootstrap inference in model averaging for predictive regressions. We first show that the standard pairwise bootstrap is not valid in the context of model averaging. This common bootstrap approach induces a bias-related term in the bootstrap variance of averaging estimators. We then propose and justify a fixed-design residual-based bootstrap resampling approach for model averaging. In a local asymptotic framework, we show the validity of the bootstrap in estimating the variance of a combined forecast and the asymptotic covariance matrix of a combined parameter vector with fixed weights. Our proposed method preserves non-parametrically the cross-sectional dependence between different models and the time series dependence in the errors simultaneously. The finite sample performance of these methods is assessed via Monte Carlo simulations. We illustrate our approach using an empirical study of the Taylor rule equation with 24 alternative specifications.

We examine forecast accuracy and efficiency of the Social Security Administration's projections for cost rate, trust fund balance, trust fund ratio made during 1980-2020 with horizons up to 95 years. We find that the deterioration in the accuracy of the forecasts during 2010's has reversed in recent years. The level of informational inefficiency has been pervasive during 1990-2009, although it shows signs of improvement after 2010.

In this paper, we use bootstrap approach to test the null hypothesis that all forecasters in the U.S. Surveys of Professional Forecasters (SPF) have equal ability. Our bootstrap procedure captures any potential cross-sectional and serial correlation in the forecast errors while preserving the unbalanced nature of the panel data. Once we account for the presence of cross-sectional and serial correlation in the forecast errors while resampling, we find convincing evidence that some individuals really are better than others-this is in sharp contrast to the findings of D'Agostino et al. (2012).

Macroeconomic expectations of various economic agents are characterized by substantial cross-sectional heterogeneity. In this paper, we focus on expectations heterogeneity among professional forecasters. We first present stylized facts and discuss theoretical explanations for heterogeneous expectations. We then provide an overview of the empirical evidence supporting the different theories and point to directions for future research. Our literature review is complemented by empirical evidence based on the ZEW Financial Market Survey, covering the behavior of expectations heterogeneity during the recent surge in inflation in 2021 and 2022. A central finding is that differences in perceptions about the workings of the economy and heterogeneity in perceptions of the precision of new signals drive disagreement among professional forecasters. While the level of disagreement varies over the business cycle, differences in beliefs persist over time.

Two recession-derivative indicators (RDIs) have been used extensively as forecast objects in business cycle prediction, viz. (1) the target variable takes value 1 if there is a recession starting exactly at a specific horizon in the future, and (2) the target variable takes value 1 if there is a recession starting any time over a specified period in the future. Using daily yield spread as an illustrative predictor, we formally and quantitatively compare the two RDIs using the receiver operating characteristics analysis. Over 1962–2021 covering eight NBER recessions, we find that generally the second RDI, ceteris paribus, will make the the predictor better performing. However, the first RDI can generate better-looking and more useful predictions under certain scenarios, depending on forecast horizon, recession duration and time profile of signals. We also consider a semiannual chronology proposed by Peláez (J Macroecon 45:384–393, 2015) and find that its performance is in the middle of the other two. Our analysis suggests that the choice of a particular RDI should be dictated by the needs of forecast user in a particular decision making context.

We develop a new econometric model for the purpose of predicting binary outcomes based on an ensemble of predictors. The method uses the pair-copula construction (PCC) to optimally combine diverse information. As a building block of PCC, the conditional copula is permitted to depend on the conditioning variable in a nonparametric way. This is the major methodological departure from our previous work. We apply this methodology to predict US business cycle peaks 6 months ahead based on the three prominent leading indicators currently used by The Conference Board. In terms of the predictive accuracy as measured by the receiver operating characteristic curve, the proposed scheme is found to do well in comparison with some popular combination models. We have also evaluated the probability forecasts generated from these models using a battery of diagnostic tools, each of which reveals different aspects of skill of the generated forecasts.

We have studied the relationship between Receiver Operating Characteristic (ROC) curve and Precision-Recall Curve (PRC) both analytically and using a real-life empirical example of yield spread as a predictor of recessions. We show that false alarm rate in ROC and inverted precision in PRC are analogous concepts, and their difference is determined by the interaction of sample imbalance and forecast bias. We found that in cases of severe class imbalance, the forecasts need to be adequately biased to mitigate the effect of imbalancedness. The mix of values of precision and recall over six sub-samples show that the predictive power of the spread has not deteriorated in recent decades, provided the optimum values of threshold are used. Using PRC, we quantify the extent to which ROC could be exaggerating the true predictive value of the yield curve in predicting recessions.

The role of education and race in explaining disparities in health-adjusted life expectancy (HALE) for Americans aged 45–64 is examined. We compute severity-weighted prevalence of diseases with comorbidity adjustments and map the information onto 21 disabling conditions from the Health and Retirement Study over 2000–2016. The approach allows us to evaluate the importance of major disease and risk factors that explain the dynamics of life expectancy and HALE in recent years, finding that Americans have been experiencing a higher prevalence of various diseases and risk factors long before the recent decline in life expectancy in 2014.

We have argued that from the standpoint of a policy maker, the uncertainty of using the average forecast is not the variance of the average, but rather the average of the variances of the individual forecasts that incorporate idiosyncratic risks. With a slight reformulation of the loss function and a standard factor decomposition of a panel of forecasts, we show that the uncertainty of the average forecast can be expressed as the disagreement among the forecasters plus the volatility of the common shock. Using new statistics to test for the homogeneity of idiosyncratic errors under the joint limits with both T and n approaching infinity simultaneously, we show that some previously used measures significantly underestimate the conceptually correct benchmark forecast uncertainty.

We forecast New York state tax revenues with a mixed-frequency model using several machine learning techniques. We found that boosting with two dynamic factors extracted from a select list of New York and U.S. leading indicators did best to correctly update revenues for the fiscal year in direct multi-step out-of-sample forecasts. These forecasts were found to be informationally efficient over 18 monthly horizons. In addition to boosting with factors, we also studied the advisability of restricting boosting to select the most recent macro variables to capture abrupt structural changes. Since the COVID-19 pandemic upended all government budgets, our boosted forecasts were used to monitor revenues in real-time for the fiscal year 2021. Our estimates showed a drastic year-over-year decline in actual revenues by over 16% in May 2020, followed by several upward nowcast revisions that led to a recovery of −1% in March 2021, which was close to the actual annual value of −1.6%.

Even though many studies have established the existence of structural breaks and declining predictability in the relationship between GDP growth and yield spreads, business analysts continue to watch for the inversion of the spread as one of the leading indicators for recessions. We use the Receiving Operating Characteristics (ROC) approach, to reevaluate the enduring power of spread to forecast recessions, notwithstanding the temporal instabilities. We identify the value of the spread that produces the highest discriminatory power as measured by different functionals of the ROC curve e.g., the hit rate, false alarm rate, and the Youden’s index. Based on data starting from January 2, 1962, we find that the optimal threshold has drifted upwards from zero since the early 1980s, and the deteriorating power of yield spread can largely be restored once the optimal cut-off values are used to issue recession forecasts.

Following recent econometric developments, we use self-assessed general health on a Likert scale conditioned by several objective determinants to measure health disparity between non-Hispanic Whites and minority groups in the United States. A statistical decomposition analysis is conducted to determine the contributions of socio-demographic and neighborhood characteristics in generating disparities. Whereas, 72% of health disparity between Whites and Blacks is attributable to Blacks’ relatively worse socio-economic and demographic characteristics, it is only 50% for Hispanics and 65% for American Indian Alaska Natives. The role of a number of factors including per capita income and income inequality vary across the groups. Interestingly, “blackness” of a county is associated with better health for all minority groups, but it affects Whites negatively. Our findings suggest that public health initiatives to eliminate health disparity should be targeted differently for different racial/ethnic groups by focusing on the most vulnerable within each group.

This paper constructs a composite leading index for business cycle prediction based on vine copulas that capture the complex pattern of dependence among individual predictors. This approach is optimal in the sense that the resulting index possesses the highest discriminatory power as measured by the receiver operating characteristic (ROC) curve. The model specification is semi-parametric in nature, suggesting a two-step estimation procedure, with the second-step finite dimensional parameter being estimated by QMLE given the first-step non-parametric estimate. To illustrate its usefulness, we apply this methodology to optimally aggregate the 10 leading indicators selected by The Conference Board (TCB) to predict economic recessions in the United States. In terms of the discriminatory power, our method is significantly better than the Index used by TCB.