EDWARD C. PRESCOTT

EDWARD C. PRESCOTT

Modern macroeconomics
Edward Prescott was born in 1940 in Glens Falls, New York and obtained his
BA (Maths) from Swarthmore College in 1962, his MS (Operations Research)
from Case Institute of Technology in 1963 and his PhD from
Carnegie-Mellon University in 1967. He was Assistant Professor of Economics
at the University of Pennsylvania (1966–71), Assistant Professor (1971–2),
Associate Professor (1972–5) and Professor of Economics (1975–80) at
Carnegie-Mellon University, and Regents’ Professor at the University of Minnesota
(1980–2003). Since 2003 he has been Professor of Economics at
Arizona State University.
Professor Prescott is best known for his highly influential work on the
implications of rational expectations in a variety of contexts and more
recently the development of stochastic dynamic general equilibrium theory.
He is widely acknowledged as a leading advocate of the real business cycle
approach to economic fluctuations. In 2004 he was awarded, with Finn
Kydland, the Nobel Memorial Prize in Economics for ‘contributions to
dynamic macroeconomics: the time consistency of economic policy and the
driving forces behind business cycles’. Among his best-known books are:
Recursive Methods in Economic Dynamics (Harvard University Press, 1989),
co-authored with Nancy Stokey and Robert E. Lucas Jr, and Barriers to
Riches (MIT Press, 2000), co-authored with Stephen Parente. His most
widely read articles include: ‘Investment Under Uncertainty’, Econometrica
EDWARD C. PRESCOTT
Edward C. Prescott 345
(1971), co-authored with Robert E. Lucas Jr; ‘Rules Rather Than Discretion:
The Inconsistency of Optimal Plans’, Journal of Political Economy
(1977), co-authored with Finn Kydland; ‘Time to Build and Aggregate
Fluctuations’, Econometrica (1982), co-authored with Finn Kydland; ‘Theory
Ahead of Business Cycle Measurement’, Federal Reserve Bank of
Minneapolis Quarterly Review (1986); ‘Business Cycles: Real Facts and a
Monetary Myth’, Federal Reserve Bank of Minneapolis Quarterly Review
(1990), co-authored with Finn Kydland; ‘The Computational Experiment:
An Econometric Tool’, Journal of Economic Perspectives (1996), co-authored
with Finn Kydland; and ‘Prosperity and Depression’, American Economic
Review (2002).
We interviewed Professor Prescott in Chicago, in his hotel room, on 3 January
1998, while attending the annual conference of the American Economic
Association.
Background Information
Where and when did you first study economics?
I first studied economics as a graduate student at Carnegie-Mellon in 1963,
which was then the Carnegie Institute of Technology. As an undergraduate I
initially started out as a physics major – back then it was the Sputnik era and
that was the glamorous field. I had two boring laboratory courses, which I
didn’t enjoy, so I transferred into math.
What was it about economics that attracted you?
Having transferred from physics to math I first considered doing applied math
– I got my degree in operations research. Then I went to an interdisciplinary
programme and it seemed to me that the smartest, most interesting people were
doing economics. Bob Lucas was a new assistant professor when I arrived at
Carnegie-Mellon. My mentor, though, was Mike Lovell, a wonderful person.
Apart from Bob Lucas and Mike Lovell, did any of your other teachers stand
out as being particularly influential or inspirational?
Sure. Morie De Groot, a great Bayesian statistician.
With respect to your own research which economists have had the greatest
influence?
I would say Bob Lucas. Also Finn Kydland, who was a student of mine –
perhaps my two most important papers were written with Finn [Kydland and
Prescott, 1977, 1982].
346 Modern macroeconomics
For over 20 years you have had a very productive relationship with Finn
Kydland. When did you first meet him?
My first position after leaving Carnegie-Mellon was at the University of
Pennsylvania. When I came back to Carnegie-Mellon Finn was an advanced
graduate student there, ready to work on research. We had a very small
economics programme with approximately seven faculty members and seven
students. It was a good programme where students worked quite closely with
faculty members. Bob Lucas and I had a number of joint students – unlike
Bob I didn’t scare the students [laughter].
Development of Macroeconomics
You have already mentioned that Bob Lucas was very influential on your own
thinking. Which other economists do you regard as being the most influential
macroeconomists since Keynes?
Well, if you define growth as being part of macroeconomics Bob Solow has
to be up there. Peter Diamond, Tom Sargent and Neil Wallace have also been
very influential.
What about Milton Friedman?
Well, I know Bob Lucas regards Friedman as being incredibly influential to
the research programme in the monetary area. Friedman’s work certainly
influenced people interested in the monetary side of things – Neil Wallace,
for example, was one of Friedman’s students. But I’m more biased towards
Neil Wallace’s programme, which is to lay down theoretical foundations for
money. Friedman’s work in the monetary field with Anna Schwartz [1963] is
largely empirically orientated. Now when Friedman talked about the natural
rate – where the unit of account doesn’t matter – that is serious theory. But
Friedman never accepted the dynamic equilibrium paradigm or the extension
of economic theory to dynamic stochastic environments.
You were a graduate student at a time when Keynesianism ‘seemed to be the
only game in town in terms of macroeconomics’ [Barro, 1994]. Were you ever
persuaded by the Keynesian model? Were you ever a Keynesian in those
days?
Well, in my dissertation I used a Keynesian model of business cycle fluctuations.
Given that the parameters are unknown, I thought that maybe you
could apply optimal statistical decision theory to better stabilize the economy.
Then I went to the University of Pennsylvania. Larry Klein was there – a
really fine scholar. He provided support for me as an assistant professor,
which was much appreciated. I also had an association with the Wharton
Economic Forecasting group. However, after writing the paper on ‘InvestEdward
C. Prescott 347
ment under Uncertainty’ with Bob Lucas [Econometrica, 1971], plus reading
his 1972 Journal of Economic Theory paper on ‘Expectations and the Neutrality
of Money’, I decided I was not a Keynesian [big smile]. I actually
stopped teaching macro after that for ten years, until I moved to Minnesota in
the spring of 1981, by which time I thought I understood the subject well
enough to teach it.
Business Cycles
The study of business cycles has itself gone through a series of cycles.
Business cycle research flourished from the 1920s to the 1940s, waned during
the 1950s and 1960s, before witnessing a revival of interest during the 1970s.
What were the main factors which were important in regenerating interest in
business cycle research in the 1970s?
There were two factors responsible for regenerating interest in business cycles.
First, Lucas beautifully defined the problem. Why do market economies
experience recurrent fluctuations of output and employment about trend?
Second, economic theory was extended to the study of dynamic stochastic
economic environments. These tools are needed to derive the implications of
theory for business cycle fluctuations. Actually the interest in business cycles
was always there, but economists couldn’t do anything without the needed
tools. I guess this puts me in the camp which believes that economics is a
tool-driven science – absent the needed tools we are stymied.
Following your work with Finn Kydland in the early 1980s there has been
considerable re-examination of what are the stylized facts of the business
cycle. What do you think are the most important stylized facts of the business
cycle that any good theory needs to explain?
Business cycle-type fluctuations are just what dynamic economic theory
predicts. In the 1970s everybody thought the impulse or shock had to be
money and were searching for a propagation mechanism. In our 1982
Econometrica paper, ‘Time to Build and Aggregate Fluctuations’, Finn and I
loaded a lot of stuff into our model economy in order to get propagation. We
found that a prediction of economic theory is that technology shocks will
give rise to business cycle fluctuations of the nature observed. The magnitude
of the fluctuations and persistence of deviations from trend match observations.
The facts that investment is three times more volatile than output, and
consumption one-half as volatile, also match, as does the fact that most
business cycle variation in output is accounted for by variation in the labour
input. This is a remarkable success. The theory used, namely neoclassical
growth theory, was not developed to account for business cycles. It was
developed to account for growth.
348 Modern macroeconomics
Were you surprised that you were able to construct a model economy which
generated fluctuations which closely resembled actual experience in the
USA?
Yes. At that stage we were still searching for the model to fit the data, as
opposed to using the theory to answer the question – we had not really tied
down the size of the technology shock and found that the intertemporal
elasticity of labour supply had to be high. In a different context I wrote a
paper with another one of my students, Raj Mehra [Mehra and Prescott,
1985] in which we tried to use basic theory to account for the difference in
the average returns on stock and equity. We thought that existing theory
would work beforehand – the finance people told us that it would [laughter].
We actually found that existing theory could only account for a tiny part of
the huge difference.
How do you react to the criticism that there is a lack of available supporting
evidence of strong intertemporal labour substitution effects?
Gary Hansen [1985] and Richard Rogerson’s [1988] key theoretical development
on labour indivisibility is central to this. The margin that they use is the
number of people who work, not the number of hours of those that do work.
This results in the stand-in or representative household being very willing to
intertemporally substitute even though individuals are not that willing. Labour
economists using micro data found that the association between hours
worked and compensation per hour was weak for full-time workers. Based on
these observations they concluded that the labour supply elasticity is small.
These early studies ignore two important features of reality. The first is that
most of the variation in labour supply is in the number working – not in the
length of the workweek. The second important feature of reality ignored in
these early studies is that wages increase with experience. This suggests that
part of individuals’ compensation is this valuable experience. Estimates of
labour supply are high when this feature of reality is taken into account. The
evidence in favour of high intertemporal labour supply elasticity has become
overwhelming. Macro and micro labour economics have been unified.
Many prominent economists such as Milton Friedman [see Snowdon and
Vane, 1997b], Greg Mankiw [1989] and Lawrence Summers [1986] have
been highly critical of real business cycle models as an explanation of aggregate
fluctuations. What do you regard as being the most serious criticisms
that have been raised in the literature against RBC models?
I don’t think you criticize models – maybe the theory. A nice example is
where the Solow growth model was used heavily in public finance – some of
its predictions were confirmed, so we now have a little bit more confidence in
that structure and what public finance people say about the consequences of
Edward C. Prescott 349
different tax policies. Bob Lucas [1987] says technology shocks seem awfully
big and that is the feature he is most bothered by. When you look at how
much total factor productivity changes over five-year periods and you assume
that changes are independent, the quarterly changes have to be big. The
difference between total factor productivity in the USA and India is at least
400 per cent. This is a lot bigger than if in say a two-year period the shocks
are such that productivity growth is a couple of per cent below or above
average. This is enough to give rise to a recession or boom. Other factors are
also influential – tax rates matter for labour supply and I’m not going to rule
out preference shocks either. I can’t forecast what social attitudes will be, I
don’t think anybody can – for example, whether or not the female labour
participation rate will go up.
In your 1986 Federal Reserve Bank of Minneapolis paper, ‘Theory Ahead of
Business Cycle Measurement’, you concluded that attention should be focused
on ‘determinants of the average rate of technological advance’. What
in your view are the main factors that determine the average rate at which
technology advances?
The determinants of total factor productivity is the question in economics. If
we knew why total factor productivity in the USA was four times bigger than
in India, I am sure India would immediately take the appropriate actions and
be as rich as the USA [laughter]. Of course the general rise throughout the
world has to be related to what Paul Romer talks about – increasing returns
and the increase in the stock of usable knowledge. But there is a lot more to
total factor productivity, particularly when you look at the relative levels
across countries or different experiences over time. For example, the Philippines
and Korea were very similar in 1960 but are quite different today.
How important are institutions?
Very. The legal system matters and matters a lot, particularly the commercial
code and the property rights systems. Societies give protection to certain
groups of specialized factor suppliers – they protect the status quo. For
example, why in India do you see highly educated bank workers manually
entering numbers into ledgers? In the last few years I have been reading quite
a lot about these types of issues. However, there seem to be more questions
than answers [laughter].
When it comes to the issue of technological change, are you a fan of
Schumpeter’s work?
The old Schumpeter, but not the new [laughter]. The new suggests that we
need monopolies – what the poor countries need is more competition, not
more monopolies.
350 Modern macroeconomics
In your 1991 Economic Theory paper, co-authored with Finn Kydland, you
estimated that just over two-thirds of post-war US fluctuations can be attributed
to technology shocks. A number of authors have introduced several
modifications of the model economy, for example Cho and Cooley [1995].
How robust is the estimate of the contribution of technology shocks to aggregate
fluctuations to such modifications?
The challenge to that number has come from two places. First, the size of the
estimate of the intertemporal elasticity of labour supply. Second, are technology
shocks as large as we estimated them to be? You can have lots of other
factors and they need not be orthogonal – there could be some moving in
opposite directions that offset each other or some moving in the same direction
that amplify each other. Are the shocks that big? Marty Eichenbaum
[1991] tried to push them down and came up with a 0.005 number for the
standard deviation of the total factor productivity shocks. My number is
0.007. I point out to Marty that Ian Fleming’s secret agent 005 is dead. Agent
007 survives [laughter].
How do you view the more recent development of introducing nominal
rigidities, imperfect credit markets and other Keynesian-style features into
RBC models?
I like the methodology of making a theory quantitative. Introducing monopolistic
competition with sticky prices has been an attempt to come up with a
good mechanism for the monetary side. I don’t think it has paid off as much
as people had hoped, but it is a good thing to explore.
The new classical monetary-surprise models developed in the 1970s by Lucas,
Sargent, Wallace and others were very influential. When did you first begin to
lose faith in that particular approach?
In our 1982 paper Finn and I were pretty careful – what we said was that in
the post-war period if the only shocks had been technology shocks, then the
economy would have been 70 per cent as volatile. When you look back at
some of Friedman and Schwartz’s [1963] data, particularly from the 1890s
and early 1900s, there were financial crises and associated large declines in
real output. It is only recently that I have become disillusioned with monetary
explanations. One of the main reasons for this is that a lot of smart people
have searched for good monetary transmission mechanisms but they haven’t
been that successful in coming up with one – it’s hard to get persistence out
of monetary surprises.
How do you now view your 1977 Journal of Political Economy paper,
co-authored with Finn Kydland, in which monetary surprises, if they can be
achieved, have real effects?
Edward C. Prescott 351
Finn and I wanted to make the point about the inconsistency of optimal plans
in the setting of a more real environment. The pressure to use this simple
example came from the editor – given the attention that paper has subsequently
received, I guess his call was right [laughter].
What do you regard to be the essential connecting thread between the monetary-
surprise models developed in the 1970s and the real business cycle
models developed in the 1980s?
The methodology – Bob Lucas is the master of methodology, as well as
defining problems. I guess when Finn and I undertook the research for our
1982 piece we didn’t realize it was going to be an important paper. Ex post
we see it as being an important paper – we certainly learnt a lot from writing
it and it did influence Bob Lucas in his thinking about methodology. That
paper pushed the profession into trying to make macroeconomic theory more
quantitative – to say how big things are. There are so many factors out there –
most of them we have got to abstract from, the world is too complex otherwise
– we want to know which factors are little and which are significant.
Turning to one of the stylized facts of the business cycle, does the evidence
suggest that the price level and inflation are procyclical or countercyclical?
Finn and I [Kydland and Prescott, 1990] found that in the USA prices since
the Second World War have been countercyclical, but that in the interwar
period they were procyclical. Now if you go to inflation you are taking the
derivative of the price level and things get more complex. The lack of a
strong uniform regular pattern has led me to be a little suspicious of the
importance of the monetary facts – but further research could change my
opinion.
What is your current view on the relationship between the behaviour of the
money supply and the business cycle?
Is it OK to talk about hunches? [laughter]. My guess is that monetary and
fiscal policies are really tied together – there is just one government with a
budget constraint. In theory, at least, you can arrange to have a fiscal authority
with a budget constraint and an independent monetary authority – in
reality some countries do have a high degree of independence of their central
bank. Now I’ve experimented with some simple closed economy models
which unfortunately get awfully complicated, very fast [laughter]. In some of
those models government policy changes do have real consequences – the
government ‘multipliers’ are very different from those in the standard RBC
model. Monetary and fiscal policy are not independent – there is a complex
interaction between monetary and fiscal policy with respect to debt management,
money supply and government expenditure. So I think that there is a
352 Modern macroeconomics
rich class of models to be studied and as we get better tools we are going to
learn more.
One of the main features of Keynesianism has always been the high priority
given by its advocates to the problem of unemployment. Equilibrium business
cycle theory seems to treat unemployment as a secondary issue. How do you
think about unemployment?
When I think about employment it is easy because you can go out and
measure it – you see how many hours people work and how many people
work. The problem with unemployment is that it is not a well-defined
concept. When I look at the experience of European economies like France
and Spain, I see unemployment as something to do with the arrangements
that these societies set up. Unemployment, particularly among the young, is
a social problem. Lars Ljungqvist and Tom Sargent [1998] are doing some
very interesting work on this question and that is something I want to study
more.
Given that your work has provided an integrated approach to the theory of
growth and fluctuations, should we perhaps abandon the term ‘business
cycle’ when we refer to aggregate economic fluctuations?
Business cycles are in large part fluctuations due to variations in how many
hours people work. Is that good language or not? I think I’ll leave that for you
to decide [laughter]. I’m sympathetic to what your question implies, but I
can’t think of any better language right now.
Methodology
You are known as a leading real business cycle theorist. Are you happy with
that label?
I tend to see RBC theory more as a methodology – dynamic applied general
equilibrium modelling has been a big step forward. Applied analyses that
people are doing now are so much better than they used to be. So in so far as I
am associated with that, and have helped get that started, I am happy with
that label.
Do you regard your work as having resulted in a revolution in macroeconomics?
No – I have just followed the logic of the discipline. There has been no real
dramatic change, only an extension, to dynamic economics – it takes time to
figure things out and develop new tools. People are always looking for the
revolutions – maybe some day some revolution will come along, but I don’t
think I’ll sit around and wait for it [laughter].
Edward C. Prescott 353
What role have calibration exercises played in the development of real business
cycle models?
I think of the model as something to use to measure something. Given the
posed question, we typically want our model economy to match reality on
certain dimensions. With a thermometer you want it to register correctly
when you put it in ice and in boiling water. In the past economists have tried
to find the model and that has held them back. Today people don’t take the
data as gospel; they look at how the data are collected. So it has forced people
to learn a lot more about government statistics on the economy.
How important was Lucas’s [1980a] paper on ‘Methods and Problems in
Business Cycle Theory’ in your development of the calibration approach?
It’s hard to recall exactly – I saw his vision more clearly later on. Back then I
kept thinking of trying to find the model, as opposed to thinking of economic
theory in terms of a set of instructions for constructing a model to answer a
particular question. There never is a right or wrong model – the issue is
whether a model is good for the purpose it is being used.
Kevin Hoover [1995b] has suggested that ‘the calibration methodology, to
date, lacks any discipline as stern as that imposed by econometric methods’.
What happens if you have a Keynesian and a real business cycle model which
both perform well? How do you choose between the two?
Well, let’s suppose you work within a Keynesian theoretical framework and it
provides guidance to construct models, and you use those models and they
work well – that’s success, by definition. There was a vision that neoclassical
foundations would eventually be provided for Keynesian models but in the
Keynesian programme theory doesn’t provide much discipline in constructing
the structure. A lot of the choice of equations came down to an empirical
matter – theory was used to restrict these equations, some coefficients being
zero. You notice Keynesians talk about equations. Within the applied general
equilibrium approach we don’t talk about equations – we always talk about
production functions, utility functions or people’s ability and willingness to
substitute. We are not trying to follow the physicist in discovering the laws of
motion of the economy, unlike Keynesians and monetarists. Keynesian approaches
were tried and put to a real test, and to quote Bob Lucas and Tom
Sargent [1978], in the 1970s Keynesian macroeconometric models experienced
‘econometric failure on a grand scale’.
To what extent is the question of whether the computational experiment
should be regarded as an econometric tool an issue of semantics?
It is pure semantics. Ragnar Frisch wanted to make neoclassical economics
quantitative – he talked about quantitative theoretical economics and quanti354
Modern macroeconomics
tative empirical economics, and their unification. The modern narrow definition
of econometrics only focuses on the empirical side.
Lawrence Summers [1991a] in a paper on ‘The Scientific Illusion in Empirical
Macroeconomics’ has argued along the lines that formal econometric
work has had little impact on the growth of economic knowledge, whereas the
informal pragmatic approach of people like Friedman and Schwartz [1963]
has had a significant effect. Are you sympathetic to Summers’s view?
In some ways I’m sympathetic, in others I’m unsympathetic – I think I’ll
hedge (laughter). With regard to representing our knowledge in terms of the
likelihood of different things being true, so that as we get more observations
over time we zero in on the truth, it doesn’t seem to work that way.
Growth and Development
Since the mid-1980s many eminent economists have turned their attention to
the issue of economic growth. Are we any closer to explaining why there has
been a lack of convergence between rich and poor countries?
The new growth and development literature, which was touched off by Paul
Romer [1986] and Bob Lucas [1988], is very exciting. We now know that
standards of living were more or less constant from the beginning of civilization
until the industrial revolution; then something changed. When I compare
countries in the East (China, India, Japan and so on) with those in the West
they were about the same in 1800 in terms of per capita GDP – by 1950 the
West was almost ten times richer, now it is only about four times richer. So I
do see signs of convergence. Divergence occurred when modern economic
growth started. In China, for example, the peasants were equally well off in
AD 2 as they were in 1950 – today they are a lot better off. The process of
modern economic growth started earlier in Japan – even so, they didn’t do all
that well until the post-war period. Japan’s relative position to England or the
USA in 1870 was about the same as it was in 1937. Even per capita income
growth in Africa is now taking place at the same rate as in the rich countries –
they should be growing much faster and I expect that they soon will start to
catch up. Furthermore, when you look at countries like India, Pakistan,
Indonesia and the Philippines, they are now growing faster than the rich
industrial countries. So I believe that there will be a lot of convergence over
the next 50 years, in the same way that there has been a lot of convergence
over the last 50 years – it all depends upon how you look at the data.
The endogenous growth literature has led to a reopening of the debate relating
to the role of government in promoting economic growth. What role do
you see for the government?
Edward C. Prescott 355
My interest is in the problem of the poor countries, like India. In those
countries it is important to let things happen and not protect the status quo.
For example, there are some bizarre licensing practices in India. Once things
start happening, they can happen pretty fast and there can be rapid development.
How do you account for the revival of interest in development economics?
People pushed the paradigm as far as the old tools would allow it to go. Now
a new generation has come along and has sought to advance it a little bit
further. Exciting theoretical developments as well new data sets are key to the
revival of interest. People like Kravis, and more recently Summers and Heston
[1991], have done a major service to the profession by providing new data.
General
If you were asked to teach macroeconomics to intermediate undergraduates,
how would you go about the task?
Basically I concentrate on the Solow growth model, with factors paid their
marginal product and consider the two key decisions: consumption–saving
and labour–leisure. In discussing monetary issues I follow some basic simple
intertemporal model with people holding assets – Neil Wallace and his students
have developed some pretty good material that can be used. The hard
thing about teaching macro to undergraduates is that the textbooks are not
that good – there is a need for a Paul Samuelson. Samuelson is an artist; he
brought undergraduates pretty well up to the level of the state of knowledge
in the profession. Now there is a big gap.
Most of your work has involved research which has pushed back the frontiers
of knowledge in economics. Have you ever thought about writing a basic
principles of economics textbook or an intermediate macro textbook?
Writing this type of book requires a special talent – if I had this talent I would
give it some serious thought. I don’t [laughter].
Have you ever been asked to be an economic adviser in Washington?
No [laughter]. I get too excited – you have to be calm and have the right
style. You also have to be a good actor as well as being a good economist. So
again I’ve never been tempted – maybe if I had the ability I might have been
asked.
Are you optimistic about the future of macroeconomics?
Yes – I think a lot of progress has been made and will continue to be made.
356 Modern macroeconomics
What issues or areas are you currently working on?
I always work on a variety of issues in the hope that one will break [laughter].
I’ve recently completed a couple of papers [Parente and Prescott, 1997;
Prescott, 1998]. One paper is on economic development for a monograph on
barriers to riches – I use game theory to construct an explicit model economy
where a particular set of monopoly rights can give rise to large differences in
total factor productivity. The other paper is on financial economics, considering
why there occurs such a big jump in the value of firms associated with
mergers. I also want to look more fully at the issue of the relationship and
interaction between monetary and fiscal policy I hinted at earlier.