Scand. J.

of Economics 121(3), 859–883, 2019

DOI: 10.1111/sjoe.12370

Paul Romer: Ideas, Nonrivalry, and

Endogenous Growth*

Charles I. Jones†
Stanford University, Stanford CA 94305-5015, USA
chad.jones@stanford.edu

Abstract
In 2018, Paul Romer and William Nordhaus shared the Sveriges Riksbank Prize in Economic
Sciences in Memory of Alfred Nobel. Romer was recognized “for integrating technological
innovations into long-run macroeconomic analysis”. This article reviews his prize-winning
contributions. Romer, together with others, rejuvenated the field of economic growth. He
developed the theory of endogenous technological change, in which the search for new ideas by
profit-maximizing entrepreneurs and researchers is at the heart of economic growth. Underlying
this theory, he pinpointed that the nonrivalry of ideas is ultimately responsible for the rise in
living standards over time.

Keywords: Economic growth; endogenous growth theory; ideas; nonrivalry; technical change
JEL classification: O3; O4

I. Introduction
When Paul Romer began working on economic growth in the early
1980s, a conventional view among economists (e.g., in the models
taught in graduate school) was that productivity growth could not be
influenced by anything in the rest of the economy. As in Solow (1956),
economic growth was exogenous. Other models had been developed in the
1960s, as discussed further below, but these failed to capture widespread
attention. Romer developed endogenous growth theory, emphasizing that
technological change is the result of efforts by researchers and entrepreneurs
who respond to economic incentives. Anything that affects their efforts,
such as tax policy, basic research funding, and education, for example, can
potentially influence the long-run prospects of the economy.
Romer’s fundamental contribution is his clear understanding of the
economics of ideas and how the discovery of new ideas lies at the heart of
economic growth. His 1990 paper is a watershed (Romer, 1990a). It stands

*I am grateful to Ufuk Akcigit, Pete Klenow, Per Krusell, Paul Romer, and Chris Tonetti for
helpful comments and suggestions.
†Research Associate, National Bureau of Economic Research.

C The editors of The Scandinavian Journal of Economics 2019.
860 Paul Romer: ideas, nonrivalry, and endogenous growth

as the most important paper in the growth literature since Solow’s Nobel-
recognized work. In this article, I review Romer’s prize-winning work,
putting it into the context of the surrounding literature and providing a
retrospective on how this research has led to the modern understanding of
economic growth. The remainder of this introduction seeks to distill these
insights into several pages, while the rest of the article delves more deeply
into the details.
The history behind Romer’s path-breaking 1990 paper is fascinating
and is engagingly presented by Warsh (2006). Romer had been working
on growth for around a decade. The words in his 1983 dissertation and
in Romer (1986) grapple with the topic and suggest that knowledge and
ideas are important to growth. Of course, at some level, everyone knew
that this must be true (and there is an earlier literature containing these
words). However, what Romer did not yet have – and what no research
had yet fully appreciated – was the precise nature of how this statement
comes to be true. By 1990, though, Romer had it, and it is truly beautiful.
One piece of evidence showing that he at last understood growth deeply is
that the first two sections of the 1990 paper are written very clearly, with
brilliant examples and precisely the right mathematics serving as the light
switch that illuminates a previously dark room.
Here is the key insight: ideas – designs or blueprints for doing
something or making something – are different from nearly every other
good in that they are nonrival. Standard goods in classical economics
are rival: as more people drive on a highway or require the skills of a
particular surgeon or use water for irrigation, there are fewer of these
goods to go around. This rivalry underlies the scarcity that is at the heart of
most of economics and gives rise to the fundamental theorems of welfare
economics.
Ideas, in contrast, are nonrival: as more and more people use the
Pythagorean theorem or the Java programming language or even the design
of the latest iPhone, there is not less and less of the idea to go around.
Ideas are not depleted by use, and it is technologically feasible for any
number of people to use an idea simultaneously once it has been invented.
Consider oral rehydration therapy, one of Romer’s favorite examples.
Until recently, millions of children died of diarrhea in developing countries.
Part of the problem is that parents, seeing a child with diarrhea, would
withdraw fluids. Dehydration would set in, and the child would die. Oral
rehydration therapy is an idea: dissolving a few inexpensive minerals,
salts, and a little sugar in water in just the right proportions produces
a solution that rehydrates children and saves their lives. Once this idea
was discovered, it could be used to save any number of children every
year – the idea (the chemical formula) does not become increasingly scarce
as more people use it.

C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 861

How does the nonrivalry of ideas explain economic growth? The key
is that nonrivalry gives rise to increasing returns to scale. The “standard
replication argument” is a fundamental justification for constant returns to
scale in production. If we wish to double the production of computers from
a factory, one way to do this is to build an equivalent factory across the
street and populate it with equivalent workers, materials, and so on. That
is, we replicate the factory exactly. This means that production with rival
goods is, at least as a useful benchmark, a constant returns process.
What Romer stressed is that the nonrivalry of ideas is an integral part
of this replication argument: firms do not need to reinvent the idea for a
computer each time a new computer factory is built. Instead, the same idea
(i.e., the detailed set of instructions for how to make a computer) can be
used in the new factory, or indeed in any number of factories, because it
is nonrival. Because there is constant returns to scale in the rival inputs
(the factory, workers, and materials), there is therefore increasing returns
to the rival inputs and ideas taken together: if you double the rival inputs
and the quality or quantity of the ideas, then you will more than double
total production.
Once you have increasing returns, growth follows naturally. Output per
person then depends on the total stock of knowledge; the stock does not
need to be divided up among all the people in the economy. Contrast this
with capital in a Solow model. If you add one computer, you make one
worker more productive. If you add a new idea – think of the computer
code for the first spreadsheet, or a word processor, or even the Internet itself
– you can make any number of workers more productive. With nonrivalry,
growth in income per person is tied to growth in the total stock of ideas
(i.e., an aggregate) not to growth in ideas per person.
It is very easy to get growth in an aggregate in any model, even in
a Solow model, because of population growth. More autoworkers produce
more cars. In a Solow model, this cannot sustain per capita growth because
we need growth in cars per autoworker. However, according to Romer, this
is not the case: more researchers produce more ideas, which ultimately
makes everyone better off because of nonrivalry. Throughout history –
25 years, 100 years, or even 1,000 years – the world is characterized by
substantial growth, both in the total stock of ideas and in the number of
people making them. Because ideas are nonrival, this is all that is required
for sustained growth in living standards.
Finally, the increasing returns associated with nonrivalry means that
a perfectly competitive equilibrium with no externalities will not exist
and cannot decentralize the optimal allocation of resources. Instead, some
departure is necessary. Romer emphasized that both imperfect competition
and externalities to the discovery of new ideas are likely to be important.
Imperfect competition provides the profits that incentivize entrepreneurs to

C The editors of The Scandinavian Journal of Economics 2019.
862 Paul Romer: ideas, nonrivalry, and endogenous growth

innovate. Later inventors and researchers benefit from the insights of those
who came before.
Research on economic growth has been monumentally influenced by
Romer’s contributions. While it can be difficult to adequately compensate
for the knowledge spillovers, the 2018 Nobel Prize in economics is a well-
deserved reward.
The remainder of this article is laid out as follows. Section II discusses
Romer’s early contributions to growth, especially Romer (1986). Section III
turns to the key insights of the 1990 paper. Section IV provides a brief
survey of the many research directions that this paper opened up. Finally,
Section V steps back and considers precisely how Romer’s insights lead to
an understanding of sustained exponential growth. Section VI concludes.1

II. Romer (1986) and the Explosion of Growth Research

The first key contribution that Romer made to the study of economic
growth was to remind the profession of the ultimate importance of this
topic. At a time when much of macroeconomics was devoted to studying
inflation and unemployment, Romer emphasized the centrality of questions
such as: “what determines the long-run rate of economic growth in living
standards?” This reminder came in the form of his 1983 dissertation
(Romer, 1983) and the key growth publication it led to (Romer, 1986).
The substantive contribution of that paper was to build a model in which
the long-run growth rate was determined endogenously, and to highlight
that, because of externalities, the equilibrium growth rate might be lower
than is optimal. In this way, Romer was a key founder of what came to be
known as endogenous growth theory.
While there is substantial merit to the details of this work, one of its
most important contributions lay in reintroducing economic growth as an
active field of study to the economics profession. As one piece of supporting
evidence, I searched for the topics “economic growth” and “endogenous
growth theory” in Microsoft Academic.2 Between 1973 and 1983, 49 papers
on economic growth that received more than 1,000 citations were published.
Between 1986 and 1996, the number rose more than five-fold, to 266.
Of course, there are many reasons for the explosion of research
on economic growth during this period. In addition to Romer’s early
work, Bob Lucas’ Marshall Lecture (Lucas, 1988), delivered by one of

1
The scientific background report prepared by the Prize Committee (2018) contains an excellent
overview of the contributions of both Romer and Nordhaus, and how they are linked.
2
Microsoft Academic (https://preview.academic.microsoft.com) allows you to sort by the number
of citations, while Google Scholar does not.

C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 863

Romer’s dissertation advisers at Chicago, was supremely influential, with

its elegant modeling, beautiful prose, and its reminder that we care not
only about growth rates but also about how enormous differences in living
standards can persist across countries. Another important factor was the
newly available data on cross-country economic performance. Summers
and Heston (1984, 1988) used the International Comparisons Project to
construct internationally comparable data on more than 100 countries back
to 1950. Maddison (1982) made his career by looking back even further in
time, to the year 1500, to shed light on how growth changed before and after
the Industrial Revolution. These data featured in several prominent empirical
papers, such as Baumol (1986), Barro (1991), Mankiw et al. (1992), and
Barro and Sala-i-Martin (1992), all of which complemented, inspired, and
pushed forward the economic theories that were being developed.
The confluence of new theories and new data made this an extremely
exciting subject and, for a time, it seemed like half the profession was
working on economic growth.3 One of the key theoretical developments
was the re-discovery of the “AK” growth model. The use of the word
“re-discovery” is intentional and highlights an important feature of the
growth literature: many of the insights of new growth theory were
anticipated, often to a great extent, by insights developed in the growth
literature of the 1950s and 1960s. Perhaps the best illustration of this point
is the AK model. Early versions of this model were developed by Von
Neumann (1945) and Frankel (1962), and Solow (1956) explicitly analyzes
this special case. In the 1980s, Lucas (1988), Jones and Manuelli (1990),
and Rebelo (1991) fit this mold, as does Romer (1986), albeit in a slightly
broader interpretation.
In its simplest form, the AK model can be expressed as
Yt = AKt (1)
and
K t = sYt − δKt , (2)
where A is an exogenous and constant productivity parameter and s is
an exogenous, constant investment rate. In this set-up, K is interpreted as
physical capital, but in Romer (1986) K was interpreted as knowledge, and
in Lucas (1988) it was replaced by human capital.

3
Some examples include Becker et al. (1990), David (1990), DeLong and Summers (1991),
Murphy et al. (1991), Stokey (1991), Young (1992), Easterly et al. (1993), Mulligan and Sala-i-
Martin (1993), Quah (1993), Parente and Prescott (1994), Sachs and Warner (1995), Stokey and
Rebelo (1995), Galor and Weil (1996), Jovanovic and Nyarko (1996), Olson (1996), Acemoglu
and Zilibotti (1997), Klenow and Rodriguez-Clare (1997), Nordhaus (1997), Weitzman (1998),
and Hall and Jones (1999).

C The editors of The Scandinavian Journal of Economics 2019.
864 Paul Romer: ideas, nonrivalry, and endogenous growth

Putting these equations together,

Yt
gY ≡ = s A − δ. (3)
Yt
The growth rate of the economy is endogenously determined by fundamental
parameters of the economic environment. In this example, a permanent
increase in the investment rate s4 will permanently raise the growth rate of
the economy.
The AK insight (i.e., a linear differential equation of the form K t = s AKt
could generate an endogenous exponential growth rate) was the theoretical
spark that lit a thousand lamps. It led to the development of a host of
different growth models and, together with the widespread availability of
data that allowed income comparisons across countries and over time, fed
the explosion of the new growth literature.

III. Romer (1990)

The watershed contribution highlighted by the Prize Committee is Romer
(1990a), and this paper makes three key contributions:
1. it identifies the nonrivalry of ideas as crucial to economic growth;
2. it highlights the role of profit-maximizing entrepreneurs and imperfect
competition;
3. it places the key AK linearity in the idea production function.

The Nonrivalry of Ideas

It is the first insight that is truly fundamental and leads to a deep,
intuitive understanding of economic growth. Many others before Romer had
appreciated that ideas are special; some of these will be discussed further
below. Indeed, Romer himself, in his 1986 paper, writes extensively about
knowledge. However, in the first eight pages of his 1990 paper, Romer put
the nonrivalry of ideas on center stage in a way that no other paper had
done before. Everything of ultimate importance follows from that insight.
Whereas Solow divided the world into capital and labor, Romer makes
a more basic distinction: between ideas, on the one hand, and everything
else (call them “objects”) on the other.5 Objects are the traditional goods

4
Note that s could be an investment rate in physical capital, human capital, or knowledge,
depending on the interpretation of K, and it is typically endogenized and, in turn, depends
on policy.
5
Romer (1993) introduced this “ideas versus objects” language.

C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 865

that appear in economics, including capital, labor, human capital, land,

highways, lawyers, a barrel of oil, a bushel of soybeans, or even a certain
volume of clean air. An idea is a design, a blueprint, or a set of instructions
for starting with existing objects, and transforming or using them in some
way that generates either more output or more utility. Examples include
calculus, the recipe for a new antibiotic, Beethoven’s Fifth Symphony, the
design of the latest quantum computer, and the technology that converts
my key strokes into electric signals that show up as letters on my computer
screen.
What is the distinction between ideas and objects? Objects are rival: one
person’s use of an object precludes the simultaneous use of the object by
others. In contrast, ideas are nonrival: an idea can be used simultaneously
by any number of people. If I use a machine, or a gallon of gasoline, or
an hour of a doctor’s time, others cannot use the same thing; this is the
sense in which objects are scarce. Indeed, economics is precisely the study
of the allocation of such scarce resources. But any number of people can
simultaneously use calculus and any number of assembly lines can use the
design for the latest computer chip. New ideas are scarce in that we are
always looking for better ideas, but existing ideas are not scarce in the same
way that a building or a doctor is – and this has far-reaching consequences
for economics.
This simple distinction is at the heart of Romer’s theory of economic
growth. Solow’s hypothesis was that the endogenous accumulation of
objects (capital) could explain growth. This hypothesis certainly had
empirical credibility: a large part of the difference between the world today
and the world in 1800 is the vast amount of capital we now have at our
disposal. However, what Solow showed is that the accumulation of objects
in a neoclassical setting runs into diminishing returns and cannot, by itself,
sustain exponential growth. However, ideas are different. Nonrivalry means
that the accumulation of ideas is not ruled out as a theory of growth by
Solow’s famous result.
To see why, Romer highlighted that the nonrivalry of ideas means that
production is characterized by increasing returns to scale. This reasoning
featured “precisely the right mathematics serving as the light switch”,
as mentioned above. Let A denote an index of the level of technology
(e.g., the stock of knowledge or the number of ideas) and let the vector
X denote all the other – rival – inputs into production. For example,
suppose we are producing a new state-of-the-art antibiotic. A denotes the
knowledge in the economy for making pharmaceuticals, and one piece of
that knowledge is the newly invented instructions for manufacturing the new
antibiotic. Perhaps it requires some extremely rare and special ingredients,
and can only be assembled by the most expert chemists. The ingredients,
the chemists, the laboratory equipment, etc., are all part of the vector X.

C The editors of The Scandinavian Journal of Economics 2019.
866 Paul Romer: ideas, nonrivalry, and endogenous growth

Indeed, the presence of human capital or costly training, or even inputs

in finite supply, do not change Romer’s basic insight; these are just part
of the list of rival inputs. Finally, let Y denote the number of doses of
the antibiotic that are produced. Then the basic production function for the
antibiotic is
Y = F(A, X). (4)
Given a certain quantity of objects X and the set of knowledge A, the
function F(·) tells us how many doses of the antibiotic are produced.
Now consider the properties of F(·). In particular, suppose we wish
to double the production of antibiotics. The standard replication argument
tells us that one way to do this is by doubling the objects (X): if we set
up a second lab that is identical in every way – with identical chemists,
identical collections of the rare ingredients, and identical equipment – then
we should be able to produce an identical number of doses, so that the
second lab will double production. In other words, for some λ > 1,
F(A, λX) = λY . (5)
That is, there is constant returns to scale to the objects in production. But
equation (5) makes clear that we have implicitly used the nonrivalry of
ideas in the replication argument: the new lab can use the same set of
instructions for building the new antibiotic. In particular, we do not have
to re-invent the idea. Once the instructions have been invented, they can be
used in one lab, two labs, or any number of labs simultaneously.
This means that, as long as more knowledge is useful, if we double the
objects and double the knowledge as well, we will more than double the
output:
F(λA, λX) > F(A, λX) = λY . (6)
That is, production is characterized by increasing returns to scale. Because
of the nonrivalry of ideas, there is constant returns to objects and therefore
increasing returns to objects and ideas taken together. Each idea only needs
to be invented once and then it is technologically feasible for the idea to
be used by any number of people or firms simultaneously and repeatedly.
Of course, just because it is feasible to do so does not mean that
this will always happen. Intellectual property rights, such as patents, might
restrict the use of ideas, and ideas can be forgotten so that they might need
to be reinvented, such as concrete in ancient Rome. These are important
considerations, but they do not change the essential point.
We now reach a crossroads in our narrative, and there are different
ways in which we can proceed. First, of course, one would like to know
precisely how the increasing returns that results from nonrivalry leads to the
possibility of sustained exponential growth. A loose guide to the answer is

C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 867

that Solow had constant returns to objects and therefore diminishing returns
to capital, and this is what dooms growth in a neoclassical model. Now that
we have increasing returns to objects and ideas together, it is not clear that
diminishing returns to – well, to what? – dooms us in the same way. More
care is required to make this intuition precise, and we will come back to
this absolutely critical point a little later. Before that, however, we follow
Romer (1990a) closely and discuss the role of imperfect competition and
profits in incentivizing innovation.

The Role of Imperfect Competition

With increasing returns to scale, inputs cannot each be paid their marginal
product in some perfectly competitive economy, as there is not enough
output to go around. As an application of Euler’s theorem, the constant
returns to rival objects implies that paying each object its marginal product
will exactly exhaust output, so then nothing is left to pay for ideas. As
a result, if ideas are to receive some compensation, at least one other
factor must be paid less than its marginal product. Conceptually, this means
that a perfectly competitive equilibrium with no externalities will not exist.
The growth literature developed two basic ways to handle this problem:
externalities and imperfect competition. Both imply that at least one factor
will be paid less than its (social) marginal product, and therefore the
equilibrium will not generally be efficient.
At a basic level, this point is profound. In particular, it implies that
a pure perfectly competitive allocation cannot decentralize the optimal
allocation of resources, and there is a role for institutions other than basic
property rights and the potential for distortionary taxes to improve the
allocation of resources. An unregulated economy is no longer the “best
of all possible worlds”.
Romer (1990a) imports the models of imperfect competition developed
by Dixit and Stiglitz (1977) and Ethier (1982) into growth theory, and
the result is elegant, both conceptually and from the standpoint of the
economics of the real world. A key to making these models applicable
was the recognition that ideas, while nonrival, are not pure public goods.
Recall that the definition of a pure public good is something that is both
nonrival and nonexcludable. While nonrivalry is a property of the economic
environment, excludability is a function of institutions and the decisions that
societies make. Institutions such as the patent system – or even just trade
secrets – can allow ideas to be partially excludable, at least for a certain
period of time.
Using this insight, the decentralized allocation in the Romer model
features entrepreneurial researchers who hunt for new ideas because of the
financial rewards that can be earned by innovating. Successful innovators

C The editors of The Scandinavian Journal of Economics 2019.
868 Paul Romer: ideas, nonrivalry, and endogenous growth

are awarded a patent that gives them the exclusive right to produce with
their invention. This allows them to charge a mark-up over marginal cost,
subject to imperfect competition, and to earn the profits that ultimately
serve as the carrot that motivates the search for new ideas.
Relative to the other growth models of the time, this framework
had a ring of truth that seemed to be missing from the literature:
growth occurs because of innovation from private entrepreneurs, who are
incentivized by the prospect of earning profits from their ideas. The contrast
with other models is worth appreciating. Solow (1956) obtained growth
only by assuming exogenous technological change in which productivity
improvements occur without anyone taking any actions. AK models based
on physical capital, such as Rebelo (1991), assume that capital actually does
not face any diminishing returns, at least in some sector, and there is no
mention of new technologies being developed. Similarly, Lucas (1988) has
individuals accumulating human capital, but again there are no inventions or
new technologies. Finally, the learning-by-doing models of Arrow (1962a)
and Frankel (1962) admit that technologies can improve but this is entirely
an accidental by-product of the production process itself rather than a result
of intentional efforts to innovate.
Constructing a model in which long-run growth is driven by the active
search for new ideas by entrepreneurs who are motivated by profits was the
second key contribution of Romer (1990a).

Linearity in the Idea Production Function

The third contribution of Romer (1990a) was to put the AK structure in
the idea production function. Essentially, this worked like the simple model
given earlier in Section II, where, as in Romer (1986), the “K” is interpreted
as knowledge rather than as physical capital. In fact, to avoid the confusion
between knowledge and capital, Romer (1990a) follows the notation we
have adopted above and denotes the stock of knowledge by A, following
tradition and reserving K for physical capital.
With At as the stock of ideas at date t, the flow of new ideas is denoted
A t ≡ dAt /dt. Romer assumes the production of new ideas is given by
A t = θH At At , (7)
where H At is the amount of human capital devoted to research and θ > 0
is a parameter governing the productivity of research. Notice that there are
increasing returns to H A and A in this idea production function, and Romer
uses externalities to handle the increasing returns in this case: individual
researchers are small and take the time path of At as exogenously given.
The new ideas they produce lead to positive knowledge spillovers that raise
the productivity of future researchers.

C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 869

The economy is endowed with a constant H̄ units of human capital

that can be used to produce either consumption goods or ideas. Letting
st ≡ H At /H̄ be the fraction of the stock of human capital that is devoted
to research, we can rewrite equation (7) as
A t
= θst H̄. (8)
At
Profits incentivize people to search for new ideas, and wages in
production encourage human capital to make consumption goods. The
market equilibrium then determines a value for st , which turns out to
be constant in the long run. Then equation (8) pins down the long-
run growth rate of the economy. This growth rate is a function of the
productivity parameter θ, the amount of human capital in the economy H̄,
and the equilibrium fraction of human capital allocated to research s. Taxes,
research subsidies, expiring patents, and other features of the economy can
influence the long-run growth rate by affecting s through the market for
entrepreneurs.
In this way, the three key ingredients – the nonrivalry of ideas,
the profit motive of imperfect competition, and putting the key linear
differential equation in the idea production function – combine to generate
the fundamental insights of Romer (1990a).

Looking Backwards
As I mentioned earlier in this article, it is remarkable how many of the
discoveries of new growth theory were actually re-discoveries of old growth
theory. For reasons we will leave to the historians, many of the insights from
the 1960s were forgotten by the heyday of the 1980s and 1990s. It is worth
reviewing the contributions from the 1960s in light of Romer’s 1990 paper
in order to see both the similarities and the differences. In this task, we are
helped appreciably by the fact that Romer himself is careful to discuss the
earlier literature.6
Arrow (1962b) views invention as the production of information, a
commodity he describes as having “peculiar attributes” in that information
should be available free of charge (as that is its marginal cost) although
this would provide no incentive for investment in research. Uzawa (1965)
considers optimal technical change in a setting where the growth rate of
technology is determined by the fraction of the labor force working in
education. Shell (1966) clearly recognizes the failure of models of perfect
competition to deliver optimal resource allocation in the presence of ideas

6
See also Romer (1987, 1990b, 2019).

C The editors of The Scandinavian Journal of Economics 2019.
870 Paul Romer: ideas, nonrivalry, and endogenous growth

and explicitly assumes that knowledge is a pure public good. Phelps (1966)
studies the “golden rule” for research investment that maximizes steady-
state consumption.
The two papers closest to Romer (1990a) are Nordhaus (1969a)
and Shell (1973). Nordhaus (1969a) notes the public-good characteristics
of knowledge and explicitly introduces “a multitude of temporary little
monopolies on information” based on fixed-life patents. His model
generates a steady state with exponential growth at a rate that is
proportional to the population growth rate. From the standpoint of Romer
(1990a), this was viewed as something of a limitation, though I will come
back to this point in Section IV.7
Each of the papers mentioned up until now notes that knowledge
is special, and they often speak of a public good. They also assume a
production function that exhibits increasing returns when A is included
as a factor of production, such as Y = AK α L 1−α . However, with one
exception, the papers do not clearly connect the nonrivalry of knowledge to
the increasing returns they have implicitly assumed. In fact, this particular
Cobb–Douglas specification was so common in the “exogenous A” case that
the production function was often assumed to take this form automatically,
rather than justified on first principles. It is an interesting accident that
the production function of exogenous growth models, if naively used in an
attempt to endogenize A, leads to the right destination even if the nuances
of nonrivalry are not fully appreciated. Later researchers, such as Griliches
(1979), who thought carefully about where to put ideas in a production
function, in some ways chose the natural assumption of treating knowledge
as another form of rival capital, putting the stock inside the constant returns
as in Y = Aβ K α L 1−α−β .
The clear exception to this statement is Shell (1973). Shell states that
technical knowledge is a public good and uses the replication argument
to justify constant returns to capital and labor, and therefore increasing
returns to capital, labor, and technical knowledge. He further appreciates
that this means that a perfectly competitive model cannot support innovation
because payments to conventional factors would exhaust output. Despite
this promising start, the models in Shell (1973) are less satisfying. They do
not generate stable exponential growth and are not fully worked out. Shell
is unable to provide a sharp characterization of economic growth, beyond
noting that the growth rate of productivity is projected to increase over time
under some conditions.

7
Phelps (1966) also derived the result, discussed in detail below, that long-run growth in per capita
income is driven by population growth. Judd (1985) obtains this result in an expanding-variety
set-up. The learning-by-doing models of Arrow (1962a) and Sheshinski (1967) contain a similar
result.

C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 871

The “new growth” insights of old growth theory were impressive,

culminating with Nordhaus (1969a) and Shell (1973). Nevertheless, for
reasons that remain unclear, these papers represented the end of an era of
active research on endogenous technical change rather than a beginning.8

IV. The Growth Literature after Romer (1990)

The explosion of the new growth literature was already well under way
when Romer (1990a) was being presented in conferences in 1988. In this
section, we highlight several different directions in which the literature
evolved.
Probably the most important is Schumpeterian growth theory, associated
most prominently with Aghion and Howitt (1992). Romer developed his
theory in the context of the “love of variety” models of Dixit and Stiglitz
(1977) and Ethier (1982), in which an idea is the discovery of how to
produce a new good, such as the internal combustion engine, the laser,
or a new antibiotic. In this setting, new inventions take a tiny share of
the market from each existing producer while at the same time increasing
the overall size of the market. The Schumpeterian model in Aghion and
Howitt (1992), in contrast, is based on “quality ladders” in which a new
innovation is a perfect substitute for some existing good, except it is of
higher quality or can be produced more cheaply. Other important early
contributions to Schumpeterian growth theory include Segerstrom et al.
(1990) and Grossman and Helpman (1991a,b).
In Schumpeterian growth models, innovation is associated with creative
destruction, in which the profit stream of a previous innovator is destroyed
(or stolen) by the creation of a new innovator. Firms can be created and
destroyed in the process of growth. In recent years, this dynamic has been
explored extensively, allowing the growth literature to connect to a wealth of
empirical facts from firm-level data. For examples, see Klette and Kortum
(2004), Aghion et al. (2014, 2017), Akcigit and Kerr (2018), and Atkeson
and Burstein (2019). Schumpeterian growth models also rely heavily on the
nonrivalry of ideas and the increasing returns that it implies, although this
is typically not emphasized.
Another important direction in which the growth literature expanded
was backwards in time. Romer (1986) originally noted that if we look back
over centuries rather than just decades, it appears that growth rates have

8
For example, as of January 2019, the Nordhaus paper had 297 citations and the Shell paper had
143 citations, both from Google Scholar. Nordhaus also has a book-length treatment published in
the same year (Nordhaus, 1969b) but it is focused on other questions related to inventive activity
rather than on the source of sustained exponential growth.

C The editors of The Scandinavian Journal of Economics 2019.
872 Paul Romer: ideas, nonrivalry, and endogenous growth

been increasing over time. It is interesting to ask to what extent models that
have been designed to explain growth in the past century can help us to
understand growth since the Industrial Revolution, and even before. Kremer
(1993) first documented that the prediction of Romer (1990a) that growth
rates should be increasing in the overall size of the economy (captured by
H̄ in the model and proxied by population in the data) actually holds up
very well when one looks at economic growth over thousands of years.9
Subsequent research by Galor and Weil (2000), Jones (2001), Hansen and
Prescott (2002), and Lucas (2002) explored the causes of growth over the
very long run; see the review of unified growth theory in Galor (2005) for
more details.
Much of Acemoglu’s work on the direction of technical change can
be interpreted as adding a second dimension to Romer (1990a): what if
researchers can invent two different kinds of ideas? In Acemoglu (1998),
ideas can augment either skilled labor or unskilled labor. Acemoglu (2003)
revisits the famous Uzawa theorem and asks if there is any reason why
technical change would tend to be labor-augmenting instead of capital-
augmenting. Acemoglu et al. (2012) consider environmental damage and
study a world in which firms can invent “clean” or “dirty” technologies.
Finally, Hemous and Olsen (2016) and Acemoglu and Restrepo (2018)
study the effects of automation on economic growth, unemployment, and
inequality.
The connections between economic growth and international trade have
been explored in a wide range of papers. Krugman (1979) and Grossman
and Helpman (1989, 1991a) were some of the early and very influential
papers. Romer himself explored this topic in Rivera-Batiz and Romer
(1991) and Romer (1994). More recent papers include Atkeson and Burstein
(2010), Ramondo et al. (2016), and Arkolakis et al. (2018).
In the end, this all-too-brief review has omitted many other topics and
lines of research that have been influenced by Romer’s work. When an
author has been cited more than 80,000 times, I suppose this is inevitable.

V. Romer and Sustained Economic Growth

The substantive economic question at the heart of this branch of the growth
literature is this: how do we understand sustained economic growth? How
is it, for example, that US GDP per person has grown at a relatively steady

9
Lee (1988) has an elegant model that combines a Romer-like production function for ideas
with a Malthusian dynamic for population to predict that the growth rate of technology will itself
grow exponentially over time.

C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 873

rate of 2 percent per year for something like the past 150 years?10 In this
section, I’d like to step back and provide an overview of the modern answer
to this question – or at least the answer that I find most persuasive – and
to show how this answer ties directly to Romer’s work.

Revisiting the Answer in Romer (1990)

The basic intuition behind sustained economic growth in Romer (1990a)
goes as follows. Ideas are nonrival, so output per person depends on
the overall stock of knowledge. Therefore, we need to understand why
knowledge can grow at a constant exponential rate. This question in turn
is answered by a key assumption: that the growth rate of knowledge
is proportional to the amount of research effort (i.e., H A above). A
constant amount of research effort therefore generates exponential growth
in knowledge.
Stated this way, there are two important problems with Romer’s answer.
First, it turns out that if you are willing to make the key assumption that
constant research effort generates exponential growth in knowledge, the
fact that ideas are nonrival then plays no essential role. Second, there is
substantial empirical evidence suggesting that this assumption is not a good
one.
To see the first problem, consider the following departure from Romer’s
model, taken recently by Akcigit et al. (2016). Suppose, counterfactually,
that knowledge, A, is completely rival, just like capital and labor, so that it
enters inside the constant returns to scale piece of the production function:
β
Yt = At Ktα LYt
1−α−β
, (9)

A t
= θ L At = θ s̄ A L̄, (10)
At

K t = s̄K Yt − δKt , (11)

and
LYt + L At = L̄. (12)
In this set-up, A is just like another form of rival capital, except it has
a different production function that involves a linear differential equation.
We assume that a constant fraction s̄ A of a fixed population is engaged

10
I think of the growth literature as consisting of two main branches. (1) How do we understand
frontier growth? (2) Why are some countries so much richer than others? While there is obviously
some overlap between these two questions, it is helpful to think of them as conceptually distinct.

C The editors of The Scandinavian Journal of Economics 2019.
874 Paul Romer: ideas, nonrivalry, and endogenous growth

in producing new ideas, so that knowledge grows exponentially at rate

g A ≡ θ s̄ A L̄. It is then straightforward to solve the rest of the model in order
to see that y ≡ Y /L (and consumption per person) will exhibit steady-state
growth at the rate11
βθ s̄ A L̄
gy = . (13)
1−α
In other words, if we are willing to assume that the idea production
function involves a linear differential equation, then the nonrivalry of
ideas becomes unnecessary and there is no need for increasing returns
in the goods production function. In terms of the taxonomy of Romer’s
contributions that I laid out in Section III, the presence of the third
contribution (linearity in the idea production function) renders the other two
contributions (nonrivalry of ideas and imperfect competition) inessential.
But the inverse also holds: if we do away with the third contribution, the
nonrivalry of ideas becomes much more important, a point that is explored
in detail below.12
The second problem is that there is ample evidence suggesting that
Romer’s original specification of the idea production function is, from
an empirical standpoint, flawed. As in equation (10), the original Romer
formulation states that the growth rate of productivity is proportional to the
amount of resources devoted to research. The problem with this formulation
is easy to see: productivity growth rates are relatively stable over time, while
the resources devoted to innovation show large exponential trends. Jones
(1995) made this point using aggregate evidence. Ngai and Samaniego
(2011) provide evidence at the industry level in the United States. In the
presence of the growing resources devoted to research that we observe
empirically in the past 50 or 100 years, the Romer set-up suggests that
we should see large increases in growth rates. Put differently, if one adds
population growth to Romer (1990a), then the model no longer generates
a steady-state growth rate. Instead, growth rates explode over time.

11
To see this, let k ≡ K/L and a ≡ A/L. Then y = k α a β (1− s̄K )1−α−β so that gy = αgk + βg a .
However, in the steady state, gk = gy and g a = g A , as there is no population growth. Combining
these equations gives the result.
12
The way I’ve stated this ignores another subtlety: the linear differential equation itself gives rise
to dynamic increasing returns. For example, suppose A t /At = s̄ L̄ and Yt = Aβt [(1 − s̄) L̄]1−β .
These equations can be combined to yield Yt = Aβ0 exp(β s̄ L̄)[(1 − s̄) L̄]1−β , which obviously
exhibits increasing returns to scale. So it would not be correct to say that all increasing returns
have been eliminated, and a perfectly competitive model with no externalities cannot decentralize
the optimal allocation here. On a related point, one could naturally ask of such a model: “why
does the idea production function look this way?” The nonrivalry of ideas would be one possible
answer.

C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 875

More recently, Bloom et al. (2019) have examined a host of evidence

at different levels of aggregation confirming that the Romer specification
of the idea production function is misguided. They look at a wide range of
evidence, from Moore’s law for semiconductors, to agricultural innovations,
to medical innovations that reduce cancer and heart disease mortality, to
firm-level data. Wherever they look, they find that “research productivity”
– defined as the ratio of the growth rate A t /At to research effort – is
declining rapidly. What was assumed to be constant by Romer (1990a) is,
in fact, falling sharply in the data. The typical estimate in Bloom et al.
(2019) suggests that research productivity declines at a rate of about 5
percent per year, meaning that the level of research productivity falls by
half every 12 years.
The weaknesses highlighted in this section are serious and require
remedy. However, in the next section, I show that dropping the empirically
untenable assumption that productivity growth is proportional to the level
of research addresses both problems. The augmented model highlights the
essential contribution of Romer (1990a) and provides a sharp and elegant
intuition for how sustained exponential growth is possible.

The Modern Romerian Answer

The modern version of Romer’s answer to the question of where exponential
growth comes from can be explained quite simply.13 In particular, we take
the basic Romer (1990a) set-up and change it only slightly, relaxing the
assumption that constant research effort can generate constant exponential
growth:
Yt = Atσ LYt (14)

A t −β
= θ L At At (15)
At

LYt + L At = Lt = L0 ent . (16)

L At = s̄Lt . (17)
Here, equation (14) says that output is produced by a production
function that has constant returns to objects (here just labor), and increasing
returns to objects and ideas together. The parameter σ measures the degree
of increasing returns to scale in the goods production function. Notice

13
The stylized model in this section is a simplified version of the models in Jones (1995), Kortum
(1997), and Segerstrom (1998). As noted earlier, related models can be found in Phelps (1966),
Nordhaus (1969a), Judd (1985), and Gomulka (1990).

C The editors of The Scandinavian Journal of Economics 2019.
876 Paul Romer: ideas, nonrivalry, and endogenous growth

that we are leaving out capital just to make the model as simple as
possible. Including capital does not change anything in a substantive way.
The original Romer model also distinguished between human capital and
unskilled labor, but we drop this distinction again for simplicity.
Equation (15) is the new production function for ideas. The parameter
β > 0 captures the extent to which ideas – or, more precisely, “proportional
improvements in productivity” – are becoming harder to find. Romer
(1990a) had a similar production function, but with β = 0. The evidence
discussed in the previous section suggested that β > 0 is more consistent
with the data. Bloom et al. (2019) find values of β ≈ 1/4 for Moore’s law
and semiconductor production, β ≈ 1 for firm-level data, and β ≈ 3 for the
aggregate economy.
Equation (16) is the resource constraint for labor – labor can be used to
make either goods or new ideas. For reasons that will become clear below,
we allow for population growth at an exogenous rate n.
Finally, equation (17) specifies that a constant fraction s̄ of the
population works to make ideas, so that 1− s̄ works to make goods. In richer
models (including those cited in footnote 13), this allocation is endogenized
and studied more carefully. Again, to get to the main point as quickly as
possible, it is appropriate to specify a simple rule for the allocation.
Once the model is set up this way, the beauty of Romer (1990a) emerges
naturally. First, notice that when we talk about economic growth, we are
speaking of growth in consumption per person. In the model, this is y ≡
Y /L and, from equation (14), this is given by
yt = Atσ (1 − s̄). (18)
The key insight of nonrivalry already shines in this equation: consumption
per person is proportional to the overall stock of knowledge, A, raised to
some power. This is because ideas are nonrival: each idea can raise each
person’s consumption. Contrast this with a Solow model, where yt = k tα ;
that is, consumption per person is proportional to capital per person, not to
the aggregate stock of capital. To increase the productivity of each person
in the economy, you need to give each person a computer; however, you
can increase the productivity of any number of people by inventing a single
new idea. In other words, the essential role of the nonrivalry of ideas has
already been embedded in the goods production function, exactly as in
Romer (1990a).
Taking logs and derivatives of equation (18), we see that the growth
rate of consumption per person is proportional to the growth rate of the
overall stock of ideas: gy = σg A. Therefore, the next step is to figure out
the long-run growth rate of ideas.
To do this, consider the idea production function in equation (15),
rewritten as

C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 877

A t L At
=θ β. (19)
At A t
In order for the left-hand side of this equation, A t /At , to be constant,
we need the right-hand side to be constant. This can only occur if the
numerator and the denominator on the right-hand side grow at the same
rate: gL A = βg A. Solving this for g A, we obtain
gL n
gA = A = . (20)
β β
The growth rate of the stock of ideas in the long run equals the growth
rate of research effort deflated by the extent to which ideas are becoming
harder to find, β. In the long run, research effort can only grow at the rate
of growth of the overall population.
Finally, combining this last expression with gy = σg A, we have
σn
gy = ≡ γn. (21)
β
This is the key result that explains long-run growth. In the long run, the
growth rate of consumption per person is the product of two terms, γ and
n. Here, γ measures the overall degree of increasing returns in the economy
and is itself equal to σ (the static degree of increasing returns in goods
production) deflated by β, the extent to which ideas are becoming harder to
find. The long-run growth rate is then the product of the overall degree of
increasing returns to scale and the rate at which the scale of the economy
is itself growing, n.14
Where does this result come from? It is tied intimately to the nonrivalry
of ideas. Recall that the original lesson of Romer (1990a) was that
nonrivalry implies increasing returns. However, increasing returns means
that bigger is more productive. An economy with more people will, other
things equal, have a larger number of researchers and therefore a larger
number of ideas in the long run. Because these ideas are nonrival, a larger
number of ideas raises everyone’s income, and an economy with more
people will be richer.
An implication of this statement is that if two otherwise identical
economies have different population growth rates, the economy with the
higher rate of population growth will grow faster: the growth rate of

14
In my earlier work on this topic, I have typically written the idea production function in terms
of the level change rather than in terms of the percentage change: A t = θ L At Aφt . The two
expressions are equivalent, with β ≡ 1 − φ. Here, the Romer case corresponds to β = 0, which
is then φ = 1 in my earlier work. Similarly, the overall degree of increasing returns to scale is
γ ≡ (σ/β) = σ/(1 − φ). The “φ” approach makes clear the connection to increasing returns
in that 1/(1 − φ) measures the dynamic increasing returns associated with the idea production
function. Both approaches have their merits.

C The editors of The Scandinavian Journal of Economics 2019.
878 Paul Romer: ideas, nonrivalry, and endogenous growth

researchers will be higher, so the growth rate of ideas will be higher, and
this delivers a higher growth rate of income per person.
In this way, the relaxation of the empirically problematic assumption
that β = 0 allows the crucial role of nonrivalry to come to the forefront.
Why has the United States grown at 2 percent per year for the past 150
years? According to this theory, an important part of the explanation is that
research effort has itself been growing exponentially. By any measure, we
have been throwing ever larger quantities of resources into the hunt for new
ideas. Just as more workers produce more cars, more researchers generate
more ideas. Exponential growth in research effort leads to a growing stock
of knowledge, and the fact that this knowledge is nonrival translates this
into growing income per person.
Notice that it is precisely the nonrivalry of ideas that is crucial. If you
try to replace knowledge with capital in this story, it no longer works. In
the presence of population growth, it is easy to get aggregate capital to
grow, just as we have gotten the aggregate stock of knowledge to grow.
However, because capital is rival, growth in the aggregate capital stock at
the rate of population growth does not translate into growth in capital per
person or income per person. In contrast, because knowledge is nonrival,
growth in the aggregate stock of knowledge at the rate of population growth
will cause income per person to grow.
The result that long-run growth in income per person is proportional
to the rate of population growth might well seem surprising, and perhaps
even controversial. This is a point that is still being debated in the growth
literature. It leads to many additional questions. Should Germany be richer
than Luxembourg because it has more people? Should Ghana grow faster
than China because it has a faster rate of population growth? Both these
questions highlight the fact that ideas flow across borders, so an individual
country’s economic performance is not intimately tied to its own population
(see Coe and Helpman, 1995; Eaton and Kortum, 1996; Klenow and
Rodriguez-Clare, 2005). What happens to growth if population growth
comes to an end? These and other questions are raised by this theory of
economic growth. Jones (2005) and Bloom et al. (2019) are two useful
references for the reader who would like to learn more about this general
topic.
The initial paragraphs of this section might appear superficially critical
of Romer (1990a), but I hope by now that I have conveyed that the truth is
actually the opposite. Removing the AK structure from the idea production
function allows the importance of nonrivalry to be fully appreciated. Such
AK structures were in vogue during the late 1980s, so it was natural for
Romer to include one. But this inclusion turned out to be a distraction from
the central contribution that Romer himself emphasized: the nonrivalry of
ideas is what makes sustained exponential growth possible.

C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 879

On this point, I cannot help but make a comment: the gods of

growth theory are human after all! Solow’s brilliant and enduring Nobel-
winning work fails to explain growth as an endogenous outcome. Romer’s
contribution, which will be celebrated for centuries, has a blemish. We
stand on the shoulders of giants, but it seems there are always a few more
cherries to be found among the higher branches. This is somehow fitting,
because if it were not the case, then growth itself might come to an
end.
This leads to one final substantive remark. Even in light of Romer’s
contributions, sustained exponential growth is surprising in another way: it
hinges on an idea production function in which the difficulty of finding new
ideas does not change dramatically over time. In the context of the simple
model above, β needs to be constant. There is no obvious reason why
this should be the case. Diamond mines can have rich veins interspersed
with poor ones, and, in the end, a mine contains only a finite supply of
diamonds. Similarly, it could be easy to find ideas for a while and then
harder, or vice versa. Or it could become ever easier over time until we
suddenly run out of new ideas. The shape of the idea production function
remains an intriguing subject of study.

VI. Conclusion
Beginning in the 1980s, the field of economic growth experienced a
renaissance. Insights from the 1950s and 1960s were rediscovered, but
the field, pushed forward by Romer, Lucas, Barro, Aghion and Howitt,
Grossman and Helpman, and many others, achieved new heights and opened
door after door for future research. Romer’s ultimate contribution stems
from an appreciation that ideas, because they are nonrival, are in their very
essence different from other goods. Romer was the first to truly understand
the implications of nonrivalry – and the increasing returns that it implies
– for economic growth.
We have a theory of endogenous technological change emphasizing
the role of entrepreneurs and researchers, motivated in part by the profits
associated with the discovery of new ideas. We have a world where the
fundamental theorems of welfare economics no longer hold; there is an
important role for government beyond simply providing secure property
rights and stepping aside. And we have an understanding of sustained
exponential growth, like that experienced in many countries for the past
150 years.
Romer’s contribution is essentially a beautiful theorem about the way
the world works, of a kind that is extremely rare. It will be remembered
and built upon for as long as economics exists.

C The editors of The Scandinavian Journal of Economics 2019.
880 Paul Romer: ideas, nonrivalry, and endogenous growth

References
Acemoglu, D. (1998), Why Do New Technologies Complement Skills? Directed Technical
Change and Wage Inequality, Quarterly Journal of Economics 113, 1055–1089.
Acemoglu, D. (2003), Labor- and Capital-Augmenting Technical Change, Journal of European
Economic Association 1, 1–37.
Acemoglu, D. and Restrepo, P. (2018), The Race between Man and Machine: Implications of
Technology for Growth, Factor Shares, and Employment, American Economic Review 108
(6), 1488–1542.
Acemoglu, D. and Zilibotti, F. (1997), Was Prometheus Unbound by Chance? Risk,
Diversification, and Growth, Journal of Political Economy 105, 709–751.
Acemoglu, D., Aghion, P., Bursztyn, L., and Hemous, D. (2012), The Environment and Directed
Technical Change, American Economic Review 102 (1), 131–166.
Aghion, P. and Howitt, P. (1992), A Model of Growth through Creative Destruction,
Econometrica 60, 323–351.
Aghion, P., Akcigit, U., and Howitt, P. (2014), What Do We Learn From Schumpeterian
Growth Theory?, in P. Aghion and S. N. Durlauf (eds.), Handbook of Economic Growth,
Vol. 2 North-Holland, Amsterdam, Chapter 1, 515–563.
Aghion, P., Bergeaud, A., Boppart, T., Klenow, P. J., and Li, H. (2017), Missing Growth from
Creative Destruction, NBER Working Paper 24023.
Akcigit, U. and Kerr, W. R. (2018), Growth through Heterogeneous Innovations, Journal of
Political Economy 126, 1374–1443.
Akcigit, U., Celik, M. A., and Greenwood, J. (2016), Buy, Keep, or Sell: Economic Growth and
the Market for Ideas, Econometrica 84, 943–984.
Arkolakis, C., Ramondo, N., Rodriguez-Clare, A., and Yeaple, S. (2018), Innovation and
Production in the Global Economy, American Economic Review 108 (8), 2128–2173.
Arrow, K. J. (1962a), The Economic Implications of Learning by Doing, Review of Economic
Studies 29, 153–173.
Arrow, K. J. (1962b), Economic Welfare and the Allocation of Resources for Invention, in
R. R. Nelson (ed.), The Rate and Direction of Inventive Activity, Princeton University Press,
Princeton, NJ.
Atkeson, A. and Burstein, A. (2010), Innovation, Firm Dynamics, and International Trade,
Journal of Political Economy 118, 433–484.
Atkeson, A. and Burstein, A. (2019), Aggregate Implications of Innovation Policy, Journal of
Political Economy, forthcoming (https://doi.org/10.1086/701823).
Barro, R. J. (1991), Economic Growth in a Cross Section of Countries, Quarterly Journal of
Economics 106, 407–443.
Barro, R. J. and Sala-i-Martin, X. (1992), Convergence, Journal of Political Economy 100,
223–251.
Baumol, W. J. (1986), Productivity Growth, Convergence and Welfare: What the Long-Run Data
Show, American Economic Review 76 (5), 1072–1085.
Becker, G. S., Murphy, K. M., and Tamura, R. (1990), Human Capital, Fertility and Economic
Growth, Journal of Political Economy 98, S12–S37.
Bloom, N., Jones, C. I., Van Reenen, J., and Webb, M. (2019), Are Ideas Getting Harder to Find?,
Working paper, Stanford University.
Coe, D. T. and Helpman, E. (1995), International R&D Spillovers, European Economic Review
39, 859–887.
David, P. A. (1990), The Dynamo and the Computer: An Historical Perspective on the Modern
Productivity Paradox, American Economic Review 80 (2), 355–361.
DeLong, J. B. and Summers, L. H. (1991), Equipment Investment and Economic Growth,
Quarterly Journal of Economics 106, 445–502.


C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 881

Dixit, A. K. and Stiglitz, J. E. (1977), Monopolistic Competition and Optimum Product Diversity,
American Economic Review 67 (3), 297–308.
Easterly, W., Kremer, M., Pritchett, L., and Summers, L. (1993), Good Policy or Good Luck?
Country Growth Performance and Temporary Shocks, Journal of Monetary Economics 32,
459–483.
Eaton, J. and Kortum, S. (1996), Trade in Ideas Patenting and Productivity in the OECD, Journal
of International Economics 40, 251–278.
Ethier, W. J. (1982), National and International Returns to Scale in the Modern Theory of
International Trade, American Economic Review 72 (3), 389–405.
Frankel, M. (1962), The Production Function in Allocation and Growth: A Synthesis, American
Economic Review 52 (5), 995–1022.
Galor, O. (2005), From Stagnation to Growth: Unified Growth Theory, in P. Aghion and S. N.
Durlauf (eds.), Handbook of Economic Growth, Vol. 1, North-Holland, Amsterdam, 171–293.
Galor, O. and Weil, D. N. (1996), The Gender Gap, Fertility, and Growth, American Economic
Review 86 (3), 374–387.
Galor, O. and Weil, D. (2000), Population, Technology, and Growth: From the Malthusian
Regime to the Demographic Transition, American Economic Review 90 (4), 806–828.
Gomulka, S. (1990), The Theory of Technological Change and Economic Growth, Routledge,
Oxford.
Griliches, Z. (1979), Issues in Assessing the Conribution of Research and Development to
Productivity Growth, Bell Journal of Economics 10, 92–116.
Grossman, G. M. and Helpman, E. (1989), Product Development and International Trade,
Journal of Political Economy 97, 1261–1283.
Grossman, G. M. and Helpman, E. (1991), Innovation and Growth in the Global Economy, MIT
Press, Cambridge, MA.
Grossman, G. M. and Helpman, E. (1991), Quality Ladders in the Theory of Growth, Review of
Economic Studies 58, 43–61.
Hall, R. E. and Jones, C. I. (1999), Why Do Some Countries Produce So Much More Output per
Worker than Others?, Quarterly Journal of Economics 114, 83–116.
Hansen, G. D. and Prescott, E. C. (2002), Malthus to Solow, American Economic Review 92 (4),
1205–1217.
Hemous, D. and Olsen, M. (2016), The Rise of the Machines: Automation, Horizontal Innovation
and Income Inequality, Working paper, University of Zurich.
Jones, C. I. (1995), R&D-Based Models of Economic Growth, Journal of Political Economy
103, 759–784.
Jones, C. I. (2001), Was an Industrial Revolution Inevitable? Economic Growth Over the Very
Long Run, Advances in Macroeconomics 1, Article 1.
Jones, C. I. (2005), Growth and Ideas, in P. Aghion and S. A. Durlauf (eds.), Handbook of
Economic Growth, North-Holland, Amsterdam, 1063–1111.
Jones, L. E. and Manuelli, R. (1990), A Convex Model of Economic Growth: Theory and Policy
Implications, Journal of Political Economy 98, 1008–1038.
Jovanovic, B. and Nyarko, Y. (1996), Learning by Doing and the Choice of Technology,
Econometrica 64, 1299–1310.
Judd, K. L. (1985), On the Performance of Patents, Econometrica 53, 567–585.
Klenow, P. J. and Rodriguez-Clare, A. (1997), The Neoclassical Revival in Growth Economics:
Has It Gone Too Far?, in B. S. Bernanke and J. J. Rotemberg (eds.), NBER Macroeconomics
Annual 1997, MIT Press, Cambridge, MA, 73–114.
Klenow, P. J. and Rodriguez-Clare, A. (2005), Extenalities and Growth, in P. Aghion and S.
Durlauf (eds.), Handbook of Economic Growth, Vol. 1, North-Holland, Amsterdam, 817–861.
Klette, T. J. and Kortum, S. (2004), Innovating Firms and Aggregate Innovation, Journal of
Political Economy 112, 986–1018.


C The editors of The Scandinavian Journal of Economics 2019.
882 Paul Romer: ideas, nonrivalry, and endogenous growth

Kortum, S. S. (1997), Research, Patenting, and Technological Change, Econometrica 65, 1389–
1419.
Kremer, M. (1993), Population Growth and Technological Change: One Million B.C. to 1990,
Quarterly Journal of Economics 108, 681–716.
Krugman, P. R. (1979), A Model of Innovation, Technology Transfer, and the World Distribution
of Income, Journal of Political Economy 87, 253–266.
Lee, R. D. (1988), Induced Population Growth and Induced Technological Progress: Their
Interaction in the Accelerating Stage, Mathematical Population Studies 1, 265–288.
Lucas, R. E. (1988), On the Mechanics of Economic Development, Journal of Monetary
Economics 22, 3–42.
Lucas, R. E. (2002), The Industrial Revolution: Past and Future, Lectures on Economic Growth,
Harvard University Press, Cambridge, MA, 109–188.
Maddison, A. (1982), Phases of Capitalist Development, Oxford University Press, New York.
Mankiw, N. G., Romer, D., and Weil, D. (1992), A Contribution to the Empirics of Economic
Growth, Quarterly Journal of Economics 107, 407–438.
Mulligan, C. B. and Sala-i-Martin, X. (1993), Transitional Dynamics in Two-Sector Models of
Endogenous Growth, Quarterly Journal of Economics 108, 739–774.
Murphy, K. M., Shleifer, A., and Vishny, R. W. (1991), The Allocation of Talent: Implications for
Growth, Quarterly Journal of Economics 106, 503–530.
Ngai, R. and Samaniego, R. (2011), Accounting for Research and Productivity Growth Across
Industries, Review of Economic Dynamics 14, 475–495.
Nordhaus, W. D. (1969), An Economic Theory of Technological Change, American Economic
Review 59 (2), 18–28.
Nordhaus, W. D. (1969), Invention, Growth and Welfare: A Theoretical Treatment of
Technological Change, MIT Press, Cambridge, MA.
Nordhaus, W. D. (1997), Do Real Output and Real Wage Measures Capture Reality? The
History of Lighting Suggests Not, in T. F. Bresnahan and R. J. Gordon (eds.), The Economics
of New Goods, University of Chicago Press, Chicago, IL, 29–66.
Olson, M. (1996), Big Bills Left on the Sidewalk: Why Some Nations are Rich, and Others Poor,
Journal of Economic Perspectives 10, 3–24.
Parente, S. L. and Prescott, E. C. (1994), Barriers to Technology Adoption and Development,
Journal of Political Economy 102, 298–321.
Phelps, E. S. (1966), Models of Technical Progress and the Golden Rule of Research, Review of
Economic Studies 33, 133–145.
Prize Committee (2018), Scientific Background on the Sveriges Riksbank Prize in Economic
Sciences in Memory of Alfred Nobel 2018: Economic Growth, Technological Change, and
Climate Change, Technical Report.
Quah, D. (1993), Empirical Cross-Section Dynamics in Economic Growth, European Economic
Review 37, 426–434.
Ramondo, N., Rodriguez-Clare, A., and Saborio-Rodriguez, M. (2016), Trade, Domestic
Frictions, and Scale Effects, American Economic Review 106 (10), 3159–3184.
Rebelo, S. (1991), Long-Run Policy Analysis and Long-Run Growth, Journal of Political
Economy 99, 500–521.
Rivera-Batiz, L. A. and Romer, P. M. (1991), Economic Integration and Endogenous Growth,
Quarterly Journal of Economics 106, 531–555.
Romer, P. M. (1983), Dynamic Competitive Equilibria with Externalities, Increasing Returns,
and Unbounded Growth, PhD thesis, University of Chicago (unpublished).
Romer, P. M. (1986), Increasing Returns and Long-Run Growth, Journal of Political Economy
94, 1002–1037.
Romer, P. M. (1987), Growth Based on Increasing Returns to Specialization, American Economic
Review 77 (2), 56–62.


C The editors of The Scandinavian Journal of Economics 2019.
 C. I. Jones 883

Romer, P. M. (1990a), Endogenous Technological Change, Journal of Political Economy 98,

S71–S102.
Romer, P. M. (1990b), Are Nonconvexities Important for Understanding Growth?, American
Economic Review 80 (2), 97–103.
Romer, P. M. (1993), Idea Gaps and Object Gaps in Economic Development, Journal of Monetary
Economics 32, 543–573.
Romer, P. M. (1994), New Goods, Old Theory, and the Welfare Costs of Trade Restrictions,
Journal of Development Economics 43, 5–38.
Romer, P. M. (2019), Nobel Lecture: On the Possibilty of Progress (https://paulromer.net/prize/).
Sachs, J. D. and Warner, A. (1995), Economic Reform and the Process of Global Integration,
Brookings Papers on Economic Activity 1, 1–95.
Segerstrom, P. (1998), Endogenous Growth Without Scale Effects, American Economic Review
88 (5), 1290–1310.
Segerstrom, P. S., Anant, T. C. A., and Dinopoulos, E. (1990), A Schumpeterian Model of the
Product Life Cycle, American Economic Review 80 (5), 1077–1091.
Shell, K. (1966), Toward a Theory of Inventive Activity and Capital Accumulation, American
Economic Review 56 (1/2), 62–68.
Shell, K. (1973), Inventive Activity, Industrial Organisation and Economic Growth, in J. A.
Mirrlees and N. H. Stern (eds.), Models of Economic Growth, Wiley, New York, 77–100.
Sheshinski, E. (1967), Optimal Accumulation with Learning by Doing, in K. Shell (ed.), Essays
on the Theory of Economic Growth, MIT Press, Cambridge, MA.
Solow, R. M. (1956), A Contribution to the Theory of Economic Growth, Quarterly Journal of
Economics 70, 65–94.
Stokey, N. L. (1991), Human Capital, Product Quality, and Growth, Quarterly Journal of
Economics 106, 587–616.
Stokey, N. L. and Rebelo, S. (1995), Growth Effects of Flat-Rate Taxes, Journal of Political
Economy 103, 519–550.
Summers, R. and Heston, A. (1984), Improved International Comparisons of Real Product and
its Composition: 1950–1980, Review of Income and Wealth 30, 207–219.
Summers, R. and Heston, A. (1988), A New Set of International Comparisons of Real Product
and Price Levels Estimates for 130 Countries, 1950–1985, Review of Income and Wealth 34,
1–25.
Uzawa, H. (1965), Optimum Technical Change in an Aggregative Model of Economic Growth,
International Economic Review 6, 18–31.
Von Neumann, J. (1945), A Model of General Economic Equilibrium, Review of Economic
Studies 13, 1–9.
Warsh, D. (2006), Knowledge and the Wealth of Nations: A Story of Economic Discovery,
W.W. Norton, New York.
Weitzman, M. L. (1998), Recombinant Growth, Quarterly Journal of Economics 113, 331–360.
Young, A. (1992), ‘A Tale of Two Cities: Factor Accumulation and Technical Change in Hong
Kong and Singapore, in O. Blanchard and S. Fischer (eds.), NBER Macroeconomics Annual,
MIT Press, Cambridge, MA, 13–54.


C The editors of The Scandinavian Journal of Economics 2019.