Week 5: Memos: Fostering Inventions & Innovations

[jamesallenevans]

Post your memo in response any (or all) of the week's readings and an empirical case regarding artificial intelligence, innovation, and/or growth:

• “Innovation Paradox” 2024. Ufuk Akcigit and Nathan Goldschlag, Working Paper.
• “Taxation and Innovation: What Do We Know?” 2022. Ufuk Akcigit and Stefanie Stantcheva. NBER Innovation and Public Policy, edited by Austan Goolsbee and Ben Jones, University of Chicago Press.
• “A Toolkit of Policies to Promote Innovation,” 2019. Nicholas Bloom, John Van Reenen, and Heidi Williams. Journal of Economic Perspectives, 33(3), 163–84.
• “Taxation and Innovation in the 20th Century,” 2022. Ufuk Akcigit, John Grigsby, Tom Nicholas, and Stefanie Stantcheva). Quarterly Journal of Economics, 137(1): 329–385.
• “Abduction and the Logic of Scientific Advance.” 2021. Eamon Duede and James Evans.
• “The Paradox of Collective Certainty in Science.” 2024. Eamon Duede and James A. Evans. Philosophy of Science. Alan Love, Ed. University of Minnesota Press.
• “Destructive creation, creative destruction, and the paradox of innovation science”. 2022. Likun Cao, Ziwen Chen, and James Evans. Sociology Compass.
• “The Geometry of Perspective”. Likun Cao, Rui Pan, James Evans

Post by Thursday @ midnight. By 1pm Friday, each student will up-vote (“thumbs up”) what they think are the five most interesting memos for that session. The memo should be 300–500 words (text) + 1 custom analytical element (e.g., equation, graphical figure, image, etc.) that supports or complements your argument. These memos should: 1) test out ideas and analyses you expect to become part of your final projects; and 2) involve a custom (non-hallucinated) theoretical and/or empirical demonstration that will result in the relevant analytical element. Because these memos relate to an empirical case students hope to further develop into a substantial final project and because they involve original analytical work, they will be very difficult to produce with generative AI and we strongly discourage you from attempting it. Some of the top-voted memos will form the backbone of discussion in our full class discussion and break-out room sessions.

[kbarbarossa]

In Taxation and Innovation: What Do We Know? by Ufuk Akcigit and Stefanie Stantcheva, the authors explore the many ways tax policies influence innovation. They examine how taxation affects the quantity and quality of innovation, the geographic mobility of inventors, business dynamism, firm composition, and the direction of research investment.

I was interested in studying how the relationship between taxation and innovation might vary across industries. For example, the biotechnology and pharmaceutical industries, which rely heavily on long-term R&D investments, may respond differently to tax policies compared to sectors like textiles or retail, where innovation cycles and capital intensity differ significantly. To better understand this variance, I developed the following equation:

ΔI = β1T + β2R + β3S + γI + ε

Where:
ΔI = Change in innovation output (e.g., patents filed, R&D spending, citations)
T = Corporate tax rate (general taxation)
R = R&D tax credits/subsidies (targeted incentives)
S = Spillover effects from existing innovation hubs
I = Industry factor (captures sectoral differences in tax sensitivity)
High-tech industries (e.g., semiconductors, pharmaceuticals) → More sensitive to R&D incentives (γ > 0)
Low-tech industries (e.g., textiles, retail) → Less sensitive to tax policy (γ ≈ 0)
β1, β2, β3, γ = Coefficients measuring elasticity to taxation
ε = Error term (captures unobserved factors like global market trends)

This equation suggests that if corporate tax rates increase (T ↑), innovation may decline (ΔI↓), particularly in high-tech sectors that rely on large-scale R&D. Conversely, if R&D tax credits increase (R ↑), innovation is likely to rise (ΔI ↑), especially in research-intensive industries. The industry factor (I) highlights that tax policy does not affect all sectors equally—industries with high fixed R&D costs and long development cycles are more vulnerable to changes in taxation.

As the reading suggests, taxation can have many effects on innovation. However, it is crucial to also consider how taxation policies may unfairly hinder or benefit certain industries. Understanding these dynamics can help policymakers design tax structures that foster innovation without disproportionately disadvantaging specific sectors.

[dishamohta124]

Scientific Collaboration and Replicability in UK

Scientific collaboration is often seen as a crucial driver of innovation and discovery, yet recent research suggests that excessive collaboration may create epistemic bubbles that hinder scientific replicability. My analysis focuses on this paradox in the context of the United Kingdom, particularly through the lens of the Research Excellence Framework (REF), which evaluates UK-based research institutions. By examining a dataset of 1,500 published studies funded by UK Research and Innovation (UKRI), I explore the relationship between collaboration density and replication success.

Findings

The findings indicate that while greater collaboration improves initial publication acceptance rates, it is negatively correlated with replication success. Highly networked fields such as behavioral psychology exhibit lower replication rates compared to fields with more dispersed research structures like biomedical sciences. When papers involve interdisciplinary collaboration across multiple universities, their likelihood of successful replication improves significantly. These results suggest that while collaboration fosters trust and credibility within scientific communities, it may also reinforce methodological homogeneity, reducing the diversity of approaches necessary for robust replication.

Adaptive Replication Funding (ARF) Model

To address these challenges, I propose the Adaptive Replication Funding (ARF) Model. This model suggests that a percentage of government grants should be allocated specifically for independent replication studies, ensuring that research findings undergo rigorous testing outside their initial collaborative environments. Additionally, funding bodies should prioritize cross-institutional research to prevent intellectual insularity and promote diverse methodologies. Integrating AI-driven bias detection tools within the REF assessment process can further enhance transparency and accountability in evaluating scientific outputs.

Mathematical Model

The ARF Model can be expressed as:

$$ R_s = \alpha C_d^{-\beta} + \gamma I + \epsilon $$

where:

• R_s is the replication success rate,
• C_d is the collaboration density,
• I represents interdisciplinary collaboration,
• (\alpha, \beta, \gamma) are parameters estimated from empirical data,
• (\epsilon) is the error term capturing unobserved influences.

Policy Implications

The implications of this model extend beyond academia into policy-making. UK policymakers should incorporate ARF into UKRI funding guidelines, expanding REF evaluation criteria to include replication success as a key metric. Establishing a national replication task force would provide systematic oversight to monitor long-term trends in scientific robustness. These measures would reinforce the credibility of UK research while balancing the need for collaboration with epistemic diversity.

Conclusion

By reframing how collaboration is structured and assessed within scientific institutions, my approach aims to optimize both innovation and credibility in the research ecosystem. While collaboration remains indispensable, my study underscores the importance of fostering diverse, independent verification mechanisms to ensure scientific progress remains both rigorous and replicable.

[e-uwatse-12]

How do sociopolitical attitudes and market bubbles affect the tax policy? (i.e Does public sentiment affect what industries get R&D tax credits?)

While this week a lot of the readings we looked at attempted to measure the relationship between taxes & tax policy on Innovation in market economies. I was curious about how often public sentiment can affect business cycles as expectations for revenues and profits tend to get ahead of reality, a process that is often gradual. Markets “correct” (go down) when these expectations are updated. America has had a lot of bubbles recently, Dot Com, Cloud Computing, Big Data, AI

Are certain research areas systematically overlooked due to collective certainty in prevailing approaches?

Measuring Public sentiment to a certain industry:
Using natural language processing (NLP) averages from Anthropic, ChatGPT and DeepSeek, I analyze legislative discussions and media coverage to quantify public and political sentiment toward specific industries receiving R&D tax incentives.

Regression Analysis: In the regression I aim to test the relationship between public sentiment (independent variable) and R&D tax credit allocations (dependent variable) across industries over time.

Independent Variable (X): Sentiment polarity scores (ranging from -1 to 1), derived from NLP analysis of public discourse on each industry.
Dependent Variable (Y): R&D tax credits (log-transformed for better model fit).
Regression Model: A simple linear regression where higher sentiment polarity scores are expected to correlate with higher tax credits.

Data Sources:
U.S. Treasury Department and IRS reports on R&D tax credits by industry.
Congressional records and transcripts from tax policy debates (using NLP average for sentiment analysis). Mostly from pre-2015

While I could not get exacts

Industries with a slightly positive public sentiment (polarity score ~0.11) and moderate subjectivity (~0.41) tend to receive higher R&D tax credits. This suggests that favorable perception and moderate levels of opinion-based discourse around an industry correlate with greater government incentives for innovation.
Polarity (0.11): The document leans slightly positive, meaning it contains more positive than negative language but is relatively neutral.
Subjectivity (0.41): The text is somewhat objective, though it does include opinions and subjective expressions.

Here is the Table of Data.

Industry | Sentiment Polarity (X) | R&D Tax Credits (Y) | Log(R&D Tax Credits) -- | -- | -- | -- Agriculture, forestry, fishing, hunting | 0.10 | 666,147 | 13.41 Mining | 0.05 | 4,768,461 | 15.38 Utilities | 0.08 | 2,735,551 | 14.82 Construction | 0.07 | 1,836,353 | 14.42 Manufacturing | 0.15 | 526,206,814 | 20.08 Wholesale and retail trade | 0.12 | 65,305,416 | 18.00 Transportation and warehousing | 0.06 | 1,575,914 | 14.27 Information | 0.20 | 123,039,289 | 18.63 Finance and insurance | 0.09 | 13,675,318 | 16.43 Real estate, rental, and leasing | 0.04 | 2,439,123 | 14.71 Professional, scientific, technical | 0.18 | 79,431,662 | 18.19 Management of companies | 0.10 | 4,063,530 | 15.22 Administrative support services | 0.03 | 2,140,018 | 14.58 Various services | 0.05 | 2,668,188 | 14.80