HIP, RIP, and the robustness of empirical earnings processes
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
The dispersion of individual returns to experience, often referred to as heterogeneity of income profiles (HIP), is a key parameter in empirical human capital models, in studies of life‐cycle income inequality, and in heterogeneous agent models of life‐cycle labor market dynamics. It is commonly estimated from age variation in the covariance structure of earnings. In this study, I show that this approach is invalid and tends to deliver estimates of HIP that are biased upward. The reason is that any age variation in covariance structures can be rationalized by age‐dependent heteroscedasticity in the distribution of earnings shocks. Once one models such age effects flexibly the remaining identifying variation for HIP is the shape of the tails of lag profiles. Credible estimation of HIP thus imposes strong demands on the data since one requires many earnings observations per individual and a low rate of sample attrition. To investigate empirically whether the bias in estimates of HIP from omitting age effects is quantitatively important, I thus rely on administrative data from Germany on quarterly earnings that follow workers from labor market entry until 27 years into their career. To strengthen external validity, I focus my analysis on an education group that displays a covariance structure with qualitatively similar properties like its North American counterpart. I find that a HIP model with age effects in transitory, persistent and permanent shocks fits the covariance structure almost perfectly and delivers small and insignificant estimates for the HIP component. In sharp contrast, once I estimate a standard HIP model without age‐effects the estimated slope heterogeneity increases by a factor of thirteen and becomes highly significant, with a dramatic deterioration of model fit. I reach the same conclusions from estimating the two models on a different covariance structure and from conducting a Monte Carlo analysis, suggesting that my quantitative results are not an artifact of one particular sample.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it