Pharma R&D Productivity

Interactive model companion to R&D Productivity and the Evolving Biopharmaceutical Ecosystem (Bonabeau, Schacht & Paul, 2026). Reproduces the Paul-style pipeline funnel and NPV framework with updated phase-specific cost, cycle-time, and p(TS) estimates. Cost models L (lower-bound, ASPE) and H (large-pharma, DiMasi) combine with p(TS) models Pe (Citeline, pessimistic) and Op (BIO/QLS, optimistic) to give four baselines (L/Op, L/Pe, H/Op, H/Pe), alongside AI Scenario A (near-term), AI Scenario B (aspirational AI-native), China cycle-time reduction, and the AI A + China combination. Two cautionary scenarios (C1 and C2) carry the argument through to its ironic conclusion: the productivity gains free resources for more WIP, amplified by Chinese capacity, so more products launch and herd into the therapeutic areas that look most lucrative — and because a crowded area is well understood both scientifically and by regulators, herding improves cost, cycle time and p(TS) again, compressing V severely. They are modelled on the V/NPV tab. The discount rate auto-sets per model (L→11%, H→9%) and stays editable; costs in 2025 USD. A therapeutic-area selector layers indication-specific p(TS) offsets and Phase I–III cost multipliers on top of any scenario. You can edit every assumption directly, or upload your own parameter set from an Excel file.

Global controls

Applies TA-specific p(TS) offsets and Phase I–III cost multipliers on top of any scenario.
Auto-set per scenario (L→11%, H→9%); editable.

The Pipeline tab uses the selected scenario/modality to redraw the Paul-style funnel picture. The NPV tab uses the same assumptions plus the commercial lifecycle tables below.

Load your own parameters

Upload an Excel workbook to replace the scenario, modality, commercial, and V-case assumptions with your own. The easiest way to get the exact format is to download the template below, which contains the current values, edit it, and upload it back. Sheets that are missing or empty are simply left unchanged.

Pipeline picture

Dynamic reproduction of the notebook-style pipeline figure. Beige phases are discovery/preclinical; blue phases are clinical/regulatory.

Selected scenario summary

Phase-level computed results

Tornado sensitivity plots

Each bar varies one parameter at a time around the selected scenario and modality, while all other inputs stay fixed. Costs and cycle times are varied by a relative percentage. p(TS) is also varied by a relative percentage and capped to stay between 0.0001 and 0.999.

Tornado plot — capitalized cost at launch
Tornado plot — PV of R&D cost at T2H

Editable R&D phase values for the selected scenario

These are the scenario’s base phase values before modality-specific adjustments. Editing a value immediately redraws the pipeline figure and recalculates all NPV tables. Biologic-specific adjustments are controlled in the next table.

Editable modality-specific R&D adjustments

Biologic defaults: Phase III cost +80%; p(TS) +5 pp Phase I, +7 pp Phase II, +6 pp Phase III, +5 pp Submission-to-Launch. All cells are editable.

Editable therapeutic-area modifiers

Applied on top of the selected scenario and modality. Δp(TS) values are added to phase p(TS); cost multipliers scale Phase I–III cost per WIP. The first row is a no-op average. All cells are editable, and the selected therapeutic area drives every figure and table in the app.

Current scenario NPV summary

Program-start NPV versus peak annual gross profit
Commercial gross-profit lifecycle for selected V
Gross profits are plotted on the absolute program timeline, where year 0 is Target-to-Hit initiation.

Herding and value compression — the productivity paradox

AI and Chinese capacity improve C, T and p(TS). Cheaper, faster, more reliable programmes free resources for more WIP, so more products launch — and they launch into whichever therapeutic area currently looks most lucrative. Crowding then improves C, T and p(TS) again, because a well-trodden area is understood both scientifically and by regulators. That second-order gain is what makes the herd worse, not better. The casualty is V.

In the cautionary scenarios, the V you select stays the perceived opportunity: the peak annual gross profit the product would earn as the sole entrant, which is precisely what attracts the crowd. The realised peak is V × Φ, where Φ combines sublinear category expansion, an order-of-entry share decay, and intra-class net-price erosion. Crowding also ends economic exclusivity before the legal clock runs out: the differentiation window caps the protected period, after which gross profit erodes to a lower residual floor, faster.

NPV versus the number of products crowding the area
Realised share of the perceived opportunity, by order of entry
Φ for each rank at the scenario’s own number of entrants. Rank 1 is the first product to launch in the area.

Editable herding parameters

Only scenarios carrying a herding block appear here; every other scenario runs the unmodified V model (Φ = 1). Setting entrants and rank to 1 reproduces the uncrowded result exactly.

The paradox, decomposed

Each cautionary scenario is built in two steps on top of a productivity base. Step 2 adds the R&D gains that herding itself produces — this is the best R&D performance anywhere in the model. Step 3 adds the value compression those same gains cause. The R&D column keeps improving while the NPV column collapses; that gap is the paradox. Computed at the selected modality, V and discount rate.

NPV table by scenario and modality

The peak-gross-profit columns are editable in the “Editable V cases” table below. For the cautionary scenarios the column headings are the perceived V; the realised peak is V × Φ.

Break-even V

Editable commercial lifecycle values

Protection ends at the later of patent expiry or regulatory exclusivity expiry. Editing these values immediately updates all curves and tables.

Editable V cases

These values set the columns in the NPV table and the x-values in the NPV chart.

Protection-window check

Equations implemented

R&D model

WIP_i = 1 / Π_{j=i..N} p_j
OOP_i = WIP_i · C_i
m_i = Σ_{k<i} T_k + T_i/2
PV0_R&D = Σ_i OOP_i / (1+r)^{m_i}
CAP_launch = Σ_i OOP_i · (1+r)^{L-m_i}
PV0_R&D = CAP_launch / (1+r)^L

Commercial model

U(a) = min(1, (1-exp(-k·a))/(1-exp(-k·t_peak)))
T_protect = min( max(T_patent, L + E_regulatory), L + D )
E(t) = 1, t ≤ T_protect
E(t) = F + (1-F)exp(-λ(t-T_protect)), t > T_protect
GP(t;V)=Φ·V·U(t-L)·E(t)·1_{t≥L}
NPV0(V)=∫ GP(t;V)/(1+r)^t dt - PV0_R&D

Herding / value compression

Φ(N,n) = N^β · s(n,N) · (1-π)^{N-1}
s(n,N) = φ^{n-1} / Σ_{j=1..N} φ^{j-1}
Φ(1,1) = 1 for all β, φ, π
D = differentiation window (years after launch)
V* = PV0_R&D / PV0_GP(V=1) ∝ 1/Φ

Terms

N products launched into the crowded area; n this programme’s order of entry. β category-expansion exponent — the area grows with more entrants but sublinearly (β = 0 is a fixed pie, β = 1 no crowding). φ rank share decay, giving earlier entrants a durable share advantage. π net-price give-up per additional entrant, applied across the class. D years of economic differentiation, which caps legal protection. Break-even V scales as 1/Φ, so the perceived opportunity needed to justify a programme rises in direct proportion to how crowded the area becomes.

Export