Collection:
Growth × Asset Pricing

17 papers
Written by Opus 5.5
Table of Contents

Seventeen papers on where long-run growth meets asset prices, each with its own digest one click away. The list grew out of one paper, Franceschini’s (2026) The innovation long-run risk component, and is organised around three questions. Origins: what generates persistent fluctuations in expected productivity and consumption growth? Pricing: why do those fluctuations carry risk premia, and how would anyone identify them? Allocation: how does technological change move income and wealth between people, and how does that feed back into investment and growth? The five stages of the core path below take the questions roughly in that order, and a last group of bridges ties the finance papers to the growth and productivity literature they lean on.

The first thing the list has to face is identification. Bansal and Yaron’s long-run risk is a small, very persistent drift in expected consumption growth, and in their own calibration it explains about 4 percent of the variance of monthly consumption growth. Switch it off and the equity premium goes from 4.2 percent to slightly negative. A component can matter that much for prices and still be nearly impossible to pin down in a finite sample of consumption data, which is why Beeler and Campbell’s audit and Hansen, Heaton and Li’s measurement sit next to it. The way out is to find something observable that moves with the drift. R&D is the natural candidate: firms invest in ideas, ideas raise productivity for years, so a burst of research is news about the whole future path of growth. Kung and Schmid build that mechanism, Croce shows what a production economy needs for long-run productivity news to be priced, and Franceschini goes to the data, finds the version of R&D that behaves like the engine the models assume, and shows that markets price its shocks.

The second thing is a wedge between innovation, adoption and measured productivity. An invention is not a technology in use, and a technology in use is not yet a rise in measured tfp. Gârleanu, Panageas and Yu make invention pure luck and put everything interesting into slow adoption, and Comin and Hobijn measure adoption lags across fifteen technologies and 166 countries. Markets can capitalise the gains years before they show up in the productivity statistics, and Pástor and Veronesi show how that same lag can produce price paths that look like bubbles without anyone being irrational.

The third thing is that innovation risk is not the same as aggregate productivity risk. Whether a new technology raises aggregate output depends on how the gains propagate through the economy, which is what Hulten’s growth accounting and Baqaee and Farhi’s extension of it are about. Whether it raises the wealth of the people who own today’s stock market depends on who captures the surplus, which is the point of the creative-destruction papers. Gârleanu, Kogan and Panageas, and then Kogan, Papanikolaou and Stoffman, find that a technology shock that is good for gdp can be a bad state for incumbent shareholders, because most of the value goes to new firms and inventors the incumbents never owned. Kogan and Papanikolaou’s survey is the map of all this and is worth reading early.

The shortest path to Franceschini is Bansal–Yaron → Kung–Schmid → Croce → Franceschini, with the Kogan–Papanikolaou survey alongside.

01 · Foundations of long-run risk

Goal: understand what the model prices, and why persistent growth news is so hard to identify.

Bansal and Yaron (2004) is the founding paper: a drift in expected growth too small to see and too persistent to ignore, priced by Epstein–Zin investors who care about the difference. Its own numbers show the identification problem, since the consumption data alone cannot tell the persistent model from an i.i.d. one, and the asset market ends up being the instrument that “detects” the drift. Beeler and Campbell (2012) take the model at its word and check every prediction it makes. Price–dividend ratios do forecast future consumption volatility, as the model says, but they do not forecast stock-return volatility, which the model says they should; stock prices seem to know about future consumption risk but not about their own.

02 · The map of innovation asset pricing

Goal: see how innovation differs from an ordinary productivity shock, and why intangible capital matters.

Kogan and Papanikolaou (2019) argue that technological progress is two shocks, not one: improvements that raise the productivity of existing capital, and improvements embodied in new capital that existing firms do not own. The two have opposite effects on today’s shareholders, and a good part of the cross-section of returns is the market pricing the difference. The survey’s sharpest point is that the sign comes from a missing market — inventors cannot sell their future rents in advance — rather than from a particular choice of preferences.

03 · Endogenous long-run growth risk

Goal: trace the mechanism from R&D investment to persistent consumption growth and asset prices.

Kung and Schmid (2015) answer the question every long-run risk seminar gets: where does the drift come from? In their endogenous growth model, calibrated without reference to Bansal and Yaron’s process, the persistence of expected consumption growth comes out almost exactly where Bansal and Yaron had assumed it, and its driver is observable: R&D intensity. Croce (2014) does the production-economy groundwork and finds that the elasticity of intertemporal substitution decides the sign of the premium, not just its size; with an elasticity below one, good long-run news cuts investment and equity becomes insurance. Franceschini (2026) closes the loop with data. R&D measured the obvious ways has a unit root, which is an awkward property for the stationary engine of expected growth, so the paper finds the version that behaves like that engine, extracts its shocks, and shows that markets price them.

04 · Innovation, adoption and creative destruction

Goal: understand why technological revolutions can raise aggregate growth while destroying incumbent wealth.

Gârleanu, Panageas and Yu (2012) separate invention from adoption. Breakthroughs arrive by luck and are embodied in new capital that takes years to install, so after a breakthrough the economy is full of unexercised growth options, expected returns rise, and consumption growth follows later. That gives a link from expected returns to future consumption growth that the standard endowment models struggle to produce. Pástor and Veronesi (2009) explain the boom and bust in the stock prices of new-technology firms as rational learning: while the technology is small its uncertainty is idiosyncratic and raises prices, and once it is adopted economy-wide the same uncertainty becomes systematic and prices fall. The signature is a rising beta into the crash. Gârleanu, Kogan and Panageas (2012) make innovation a priced risk through displacement: new firms and new workers take a share of future output from those already alive, so innovation that reads as good news in aggregate data is bad news for current shareholders, and growth stocks are the hedge.

05 · Measurement and applications

Goal: connect patent values, R&D productivity and the distribution of gains to observable data.

Kogan, Papanikolaou, Seru and Stoffman (2017) value each patent by the stock-market reaction to its grant, which turns innovation into dollars. With that measure, rivals’ innovation is followed by lower output at competing firms; with citation-weighted patent counts, the same effect disappears. Creative destruction shows up only when innovation is weighted by its market value. Kogan, Papanikolaou and Stoffman (2020) use the measure to explain why good technology news can be a bad state for the typical shareholder: inventors capture most of the value of new projects, the median household falls behind, and growth stocks are insurance against being left behind by other people’s wealth creation rather than against recessions. Bloom, Jones, Van Reenen and Webb (2020) measure the input side: across chips, crops, medicine and firms, research productivity falls steadily, so constant growth needs ever more researchers.

Bridges and extensions

Four bridges, each tied to one of the questions above.

Identification. Hansen, Heaton and Li (2008) measure long-run consumption risk directly and find that the value premium lives almost entirely in long-run exposure: value and growth cash flows look alike in the short run and differ only after several years, which is exactly where the estimates are least precise.

Diffusion. Comin and Hobijn (2010) date the adoption of fifteen technologies across 166 countries and find that later inventions are adopted faster, by about 4.3 years per decade of invention date, a pattern that predates computers and postwar globalisation.

Intangibles and who captures the rents. Ai, Croce and Li (2013) treat intangible capital as a stock of investment options. Good long-run news raises the value of installed capital and makes firms postpone exercising their options, so installed capital carries the long-run risk premium and ideas are the hedge. (The note is written from the 2010 conference draft; the published version is paywalled.)

Growth accounting and networks. Hulten (1978) shows how sectoral productivity gains add up to aggregate tfp: weight each sector by its sales relative to final demand, weights that sum to more than one whenever there are intermediate inputs, because those linkages magnify technical change. Baqaee and Farhi’s Beyond Hulten’s Theorem, digested for the geoeconomics list, shows that beyond first order the network and the elasticities matter after all. Aghion and Howitt (1992) put creative destruction into an endogenous growth model: each innovation destroys the rents of the one before, so expected future research discourages research today, and laissez-faire growth can even be too fast.

The open question

Read together, the list suggests a question none of the papers answers: can financial markets tell apart innovations that raise aggregate productivity from innovations that mainly move rents between firms and factors? Imagine two technologies. Technology A raises tfp broadly, and its gains spread across sectors, workers and consumers. Technology B makes a few firms very profitable by displacing incumbents, with modest gains in aggregate productivity. Both could produce large stock-market revaluations. Their implications for consumption growth, investment, wages and long-run risk would be very different. The papers here supply the pieces for telling them apart — market-based patent values, firm-level equity responses, input–output exposures and subsequent productivity growth — but separating news about aggregate technological progress from news about who will own the profits is still open.