# US Nonimmigrant Visa Holder Estimates & Overstay Modeling (Dec 2025)
## Primary Official Sources
- DHS OHSS: *Population Estimates for Nonimmigrants Residing in the United States* (latest FY 2024, Aug 2025)
- DHS: *Entry/Exit Overstay Report* (latest FY 2023)
- DOS: *Report of the Visa Office* (FY 2024 full, FY 2025 monthly)
- USCIS: Quarterly petitions approved & Employer Data Hub
## Types of Visas
1. Diplomatic & Official
- A-1
- A-2
- A-3
- G-1
- G-2
- G-3
- G-4
- G-5
- NATO-1 to NATO-7
2. Business & Tourism
- B-1
- B-2
3. Transit & Crew
- C-1
- C-1/D
- C-2
- C-3
- D-1
4. Treaty Trader/Investor
- E-1
- E-2
- E-2C
- E-3
5. Students
- F-1
- F-2
- F-3
- M-1
- M-2
- M-3
6. Exchange Visitors
- J-1
- J-2
- Q-1
- Q-2
- Q-3
7. Specialty Occupations
- H-1B
- H-1B1
- H-1C
8. Temporary Workers
- H-2A
- H-2B
- H-3
- H-4
9. Intracompany Transferees
- L-1A
- L-1B
- L-2
10. Media & Journalists
- I
11. Fiancé(e) & Family
- K-1
- K-2
- K-3
- K-4
- V-1
- V-2
- V-3
12. Religious Workers
- R-1
- R-2
13. Extraordinary Ability
- O-1A
- O-1B
- O-2
- O-3
14. Athletes & Entertainers
- P-1A
- P-1B
- P-2
- P-3
- P-4
15. Victims of Trafficking/Crime
- T-1 to T-6
- U-1 to U-5
16. NAFTA/TN Professionals
- TN
- TD
17. Special Immigrant/Informant
- S-5
- S-6
- S-7
18.
Parent/Child of Special Immigrants
- N-8
- N-9
## 1. Cohort-Component Model (Active Holders)
$P_t = P_{t-1} + I_t - O_t - D_t + A_t$
Where:
- $P_t$ = active population at current time $t$
- $I_t$ = inflows (new issuances + admissions)
- $O_t$ = departures (estimated from prior cohorts)
- $D_t$ = adjustments to LPR or removals
- $A_t$ = extensions/renewals
Typical durations:
- H-1B: average 5.5 years (max 6)
- H-2A: average 9 months (highly seasonal)
Baseline (DHS FY 2024):
- H-1B ≈ 583 000 active
- H-2A peak ≈ 300 000 active
---
### "Estimated from prior cohorts" – Exact Meaning
In the cohort-component model, the departures term $O_t$ is **not measured directly in real time** for every individual. Instead, it is **estimated from prior cohorts** using the following method:
#### Definition
Take every group of people who entered the U.S. on the same visa class in a past year $t-k$ (a “cohort”) and calculate how many actually left on time using historical I-94 matched records. That observed departure fraction is then applied to the surviving members of each cohort still present in year $t$.
#### Mathematical Form (simplified)
$
O_t = \sum_{k=1}^{k_{\text{max}}} \left( I_{t-k} \times (1 - r_k) \times s_{t,k} \right)
$
Where:
- $I_{t-k}$ = inflows (admissions/issuances) in year $t-k$
- $r_k$ = overstay rate observed for that cohort after $k$ years
- $s_{t,k}$ = survival fraction of the cohort still legally present at time $t$ (accounts for earlier departures, adjustments, etc.)
- $k_{\text{max}}$ = maximum allowed duration of the visa class
#### Concrete Examples
##### H-1B (max 6 years)
$
O_{2025} \approx
(I_{2019} \times 0.985) +
(I_{2020} \times 0.97) + \cdots +
(I_{2024} \times 0.70)
$
(using observed departure rates per cohort year)
### H-2A (typically ≤ 10 months)
$
O_{2025} \approx I_{2024} \times (1 - 0.03) \approx I_{2024} \times 0.97
$
---
<p align="justify"><b>Note:</b> Standards for calculations are derived from the DHS/USCIS methodology to solve for “estimated from prior cohorts” in the departures term of the population formula.</p>
Referenced from *[U.S. Nonimmigrant Admissions: 2023](https://ohss.dhs.gov/sites/default/files/2024-09/2024_0906_ohss_nonimmigrant_fy2023_0.pdf)* published by the Office of Homeland Security Statistics. Authored by Aneer Rukh-Kamaa.
<div style="left: 0; width: 100%; height: 0; position: relative; padding-bottom: 60%;"><iframe src="https://docs.google.com/viewer?embedded=true&url=http://sbik.site.nfoservers.com/PDFs/Topics_A-Z/Visas-and-Census_Dec2025/US-Nonimmigrant_Admissions-2023.pdf" style="top: 0; left: 0; width: 100%; height: 100%; position: absolute; border: 0;" allowfullscreen></iframe></div>
From the Text (page 3):
- In 2023, DHS granted an estimated 132 million nonimmigrant admissions to the United States (Figure 1).
- These included 68 million admissions of nonimmigrants who were issued Form I-94, the primary focus of this report.
- Nonimmigrant admissions in 2023 increased **<span style="background:rgba(3, 135, 102, 0.2)">52 percent</span>** from 2022.
## 2. Residual Overstay Model (Visa Overstays)
$U = \sum_c (V_c \times r_c \times s_c)$
Where:
- $V_c$ = cumulative admissions/issuances in class $c$ over look-back window (FY2023)
- $r_c$ = overstay rate (DHS FY 2023: H-1B ≈ 1.5 %, H-2A ≈ 2–3 %)
- $s_c$ = survival fraction still undocumented (≈ 60–70 % per CMS)
DHS historical split: ≈ 62 % of new undocumented entrants are visa overstays.
## 3. Bayesian Updating (Uncertainty Modeling)
Prior ← DHS official estimate
Likelihood ← overstay rates + petition approvals
Posterior ← updated count + 95 % credible interval
Implementation: PyMC3 / PyMC (Python) or Stan.
## Framework Comparison
| Framework | Key Inputs | Output | Pros | Cons |
|--------------------|-------------------------------------|----------------------------|------------------------------|------------------------------|
| Cohort-Component | Issuances, durations, rates | Active holders by month | Captures dynamics | Requires monthly/quarterly data |
| Residual | Total undocumented, EWI split | Overstay undocumented | Simple stock estimate | Assumes stable rates |
| Bayesian | DHS priors + overstay likelihood | Estimate + CI | Quantifies uncertainty | Computationally heavier |
## Query Strategy (Exact Search Strings)
- `"DHS nonimmigrant population estimates FY 2025" site:dhs.gov`
- `"Entry/Exit Overstay Report FY 2023" H-1B OR H-2A site:dhs.gov`
- `"Report of the Visa Office" "Table XVI" FY 2025 site:travel.state.gov`
- `"USCIS H-1B petitions approved" FY 2025 site:uscis.gov`
## Accuracy Target
Using the above methods and official sources: 90–95 % alignment with final DHS releases when they publish (typically Q3 following fiscal year).