Equations of Motion for an Economy: Capital Deepening, Technology, and Firm Survival

The paper derives a model of economic growth based on accounting identities rather than production functions, demonstrating that capital deepening is driven by structural cost reductions rather than productivity gains in new investment, while simultaneously using firm-level exit dynamics to explain observed Zipf-like distributions.

The Gist

Imagine you are running a massive, global bakery. To understand how this bakery grows, most economists look at a "magic recipe" (called a production function) and then wonder why the bread sometimes rises more than expected.

This paper, written by Robert Nachtrieb, throws away the "magic recipe." Instead, he looks at the accounting books. He says: "Don't guess how much bread you can make; just look at how much flour you bought, how many ovens you have, and how much money is left in the cash register."

Here is the breakdown of his discovery using the bakery analogy.

1. The Two Ways to Grow: "Cheaper Ovens" vs. "Better Ovens"

The author says there are two ways a bakery can get more productive, and this is the most important part of the paper.

• The "Cheapening" Channel (The $\mu$ factor): Imagine that every year, the price of a standard oven drops. You aren't getting a better oven; you’re just getting the same oven for less money. Because they are cheaper, you can buy ten ovens instead of five. You are making more bread because you have more "stuff," even though the "stuff" isn't any smarter. This is what has driven almost all economic growth for the last 75 years. It’s like upgrading from a manual whisk to an electric mixer—it’s faster, but it’s still just a mixer.
• The "Productivity" Channel (The $\phi$ factor): This is the "Holy Grail." This is when a new oven is invented that doesn't just cost less, but actually does more per dollar. Imagine an oven that uses AI to bake bread perfectly every time without a human. This isn't just "more stuff"; it's "smarter stuff."

The Big Finding: The author looked at 75 years of data and found that the "Better Oven" ($\phi$) has been zero. Every technological revolution we’ve had—electricity, computers, the internet—has mostly just been the "Cheaper Oven" ($\mu$) effect. We’ve been getting better at making things cheaper, but we haven't fundamentally changed how much "work" a single dollar of equipment can do.

2. The "Profit Imperative": The Survival Threshold

Every bakery has a "break-even" point. If an oven costs \1,000 but only helps you make \800 worth of bread, that oven is a loser. It will bankrupt you.

The author calls this the Profit Imperative. It’s a mathematical line in the sand. If your equipment isn't producing enough profit to cover its own cost and the wages of the bakers, you are in the "danger zone."

3. Why Do Small Businesses Die? (The "Cash Martingale")

The paper also explains why some bakeries go out of business.

Imagine a small bakery that is just barely making enough money to pay the bills. They have almost no extra cash in the drawer. In the world of math, this is called a "martingale"—it means their cash level is just bouncing around randomly. Because they have no "cushion," even one bad week (a broken oven or a slow Tuesday) will drop their cash to zero.

When you combine this "danger zone" with the fact that there are always many tiny bakeries and only a few huge ones (a pattern called Zipf's Law), you get a very predictable math formula for how many businesses will fail every year. The author's math matches real-world data perfectly without him having to "tweak" the numbers to make it work.

4. The Big Prediction: The AI Test

The author ends with a massive, "falsifiable" prediction (meaning you can prove him wrong very easily).

He says: "Is AI the first time we are actually getting 'Better Ovens' ($\phi$) instead of just 'Cheaper Ovens' ($\mu$)?"

• If AI is just a tool that makes computers cheaper: The economy will keep growing at its usual, slow, steady pace.
• If AI is a "Productivity Channel" ($\phi$) breakthrough: We will see a massive, unmistakable "upward curve" in the data. It would be like the bakery suddenly doubling its output not because it bought more ovens, but because the ovens became "magically" more efficient.

The Bottom Line: We are currently in a waiting room. We are watching the data to see if AI is just a cheaper way to do old things, or if it is the spark that finally shifts the entire engine of the global economy into a higher gear.

Technical Summary

Technical Summary: Equations of Motion for an Economy

Title: Equations of Motion for an Economy: Capital Deepening, Technology, and Firm Survival
Author: Robert T. Nachtrieb (MIT Sloan School of Management)

---

1. The Problem: Moving Beyond the Production Function

Traditional macroeconomic growth theory (e.g., the Solow model) relies on production functions ($Y = AK^\alpha L^{1-\alpha}$) where Total Factor Productivity ($A$) is treated as an exogenous, unexplained residual. This approach fails to provide a mechanistic derivation of how technology and capital interact to drive growth. The paper seeks to derive the "equations of motion" for an economy using only exact accounting identities, treating growth as a dynamical system of observable intensive quantities rather than an unexplained residual.

2. Methodology: Accounting-Based Dynamical Systems

The author shifts the unit of analysis to the sector level, assuming within-sector homogeneity where firms share similar intensive quantities (capital per worker, capital productivity, and labor share) but differ in scale.

Key Variables:

• $\kappa$ (Kappa): Capital per worker (capital deepening).
• $\eta$ (Eta): Capital productivity ($Y/K$).
• $\eta_{new}$: Frontier productivity of newly purchased capital.
• $q$: Labor share ($w/y$).
• $\eta^*$: The Profit Imperative—the minimum viable capital productivity required for a firm to break even ($\Pi = 0$).

The Mathematical Framework:
The paper derives a system of four coupled Ordinary Differential Equations (ODEs) that govern the evolution of these variables. The core innovation is the decomposition of the frontier productivity equation ($\frac{d\eta_{new}}{dt}$) into two distinct physical channels:

1. The Structural Channel ($\mu$): A "cheapening" mechanism where successive generations of capital become less productive per unit of investment (diminishing returns).
2. The Productivity Channel ($\phi$): A "capability" mechanism where new capital is genuinely more productive per dollar than the existing frontier.

The model is constrained by a "sandwich invariant": $\eta^* \leq \eta_{new} \leq \eta$, ensuring that new investment is always more productive than the current average but less productive than the theoretical frontier, and always above the survival threshold.

3. Key Contributions

• First-Principles Derivation: Derives growth dynamics directly from BEA (Bureau of Economic Analysis) accounting identities without assuming a production function.
• Identification of the $\phi$ Parameter: Distinguishes between capital becoming cheaper (structural) and capital becoming smarter/more capable (productivity).
• Micro-Macro Linkage: Connects macroeconomic sector dynamics to microeconomic firm survival. It uses the "profit imperative" to show that firms near the breakeven threshold behave as a cash martingale, allowing the author to derive firm exit rates using stochastic processes.
• Falsifiable Prediction: Provides a specific, observable signature for a shift in the nature of technological progress.

4. Results

• Empirical Calibration: Using BEA 2-digit NAICS data (1998–2023) and a 75-year extended panel, the author finds that $\phi = 0$ for all identifiable sectors. This implies that every major technological wave in the postwar era (electrification, computers, the internet) has been driven by the structural channel ($\mu$)—making capital cheaper and driving growth through capital deepening ($\kappa$), rather than making capital more capable ($\phi$).
• Growth Decomposition: The data shows that real output growth ($y$) is driven almost entirely by capital deepening ($\dot{\kappa}/\kappa > 0$), while capital productivity ($\eta$) is actually declining or stagnant in most sectors ($\dot{\eta}/\eta \leq 0$).
• Firm Exit Rates: The model predicts a firm exit rate of $\sim t^{-1/2} \log t$. When convolved with a Zipf firm-size distribution, this yields an exponent of $b = 0.295 \pm 0.03$, which matches Business Dynamics Statistics (BDS) data without any free parameters.
• The "AI Test": The paper predicts that if AI represents a genuine productivity shift ($\phi > 0$), it will manifest as an upward-curving $\eta(t)$ in BEA data. A modest $\phi = 0.01$ (1% improvement in new-capital productivity) would nearly double the aggregate growth rate from 1.5% to 2.4% by 2050.

5. Significance and Policy Implications

The paper offers a profound reinterpretation of technological progress. It suggests that we have lived through a long era of "cheapening" technology, where growth is sustained by investing more capital into increasingly efficient (but not inherently more capable) tools.

Policy Implication: The author argues that attempting to accelerate the "cheapening" of capital (increasing $\mu$) may actually reduce medium-run growth because it compresses profit margins faster than capital deepening can compensate. To achieve a higher growth trajectory, the economy must shift toward the $\phi$ channel—investing in capital that is fundamentally more capable per dollar. The paper sets a clear, empirical deadline: the impact of AI on the global economy will be unambiguously visible in national accounts within one capital lifetime.