It would be difficult to find two questions of macroeconomic
theory more central to the conduct of economic policy today than these:
- Why are some nations poorer than others?
- Do changes in money supply affect economic output?
The first question is at the heart of disputes over globalization,
trade policy and international aid. The second is crucial to debates
about employment, inflation and interest rates. And although these
problems are related only indirectly to one another, they share
two characteristics: (1) Solid answers to them are essential to
sound policy; and (2) such answers have been hard to find, though
economists have worked on them for decades.
Fortunately, insights into both macroeconomic questions are now
being discovered through a rather unexpected method—the ingenious
application of price theory, a traditional domain of microeconomics.
In recent work, Minneapolis Fed senior economist Peter Klenow examines
international differences in the price ratios of investment goods
to consumption goods and finds that traditional theories of underdevelopment
aren't supported by the data. The facts suggest that the strong
correlation between levels of investment and levels of income—often
seen as a major causal relationship—is a bit of a chimera.
A closer look reveals that at domestic prices there's actually
very little correlation between investment rates and income levels.
To the extent that savings and investment rates are important in
explaining differences in development, that implies that the heart
of the problem lies not in inadequate savings, as usually proposed,
but in inefficient use of those savings. The right question, in
other words, isn't why do some nations save so little, but rather
why do they get so little from what money they do save?
And in separate research, Klenow looks at how prices change when
monetary policy changes. Here, too, he finds that the evidence is
at odds with much conventional theory. Many economists today rely
on macroeconomic models in which changes in monetary policy can
affect economic output—expanding the money supply by reducing
the federal funds rate, for instance, can boost quantities of products
demanded by consumers because not all prices adjust quickly. (With
more money in their hands, and prices stuck for a while, goes the
theory, people will demand more "sticky-price" products.)
But Klenow finds that the price and quantity changes predicted by
such sticky-price models following monetary "shocks"
don't meet the reality test—that is, the model's predictions
don't match the actual data. So either the models are flawed, the
way economists usually measure monetary shocks is flawed or both,
Klenow's exploration of price behavior in these very different contexts
is bringing new clarity to persistent questions. And such insights
aside, the most intriguing aspect of his work is how it uses elementary
price theory—and micro-level data—to delve into the
broadest and most fundamental issues of economic growth and transformation,
bridging the customary—if artificial—divide between
micro and macroeconomics.
Prices and prosperity
What causes underdevelopment? Why are some countries rich and others
poor? There are many reasons, no doubt, only some of them economic.
(See the review of The Elusive Quest
for Growth.) But for their part, economists have focused on three
sources for the variations among countries: physical capital, human
capital and total factor productivity. The last, TFP, is sort of a
black box that seems to include everything that the other two factors
don't. Human capital means levels of education and training of a country's
population, the acquired skills that people can apply toward economic
development. And physical capital is machinery, equipment and tangible
infrastructure: the trucks, roads and factories that produce and deliver
While TFP and human capital ultimately may prove to be most significant
in explaining different levels of economic development, noted Klenow
in a recent interview, economists have given more attention to physical
capital. "It's 20 percent of the story, not 100 percent, but
it's definitely more than 20 percent of the literature," he
observed. (The disproportion may be due to the fact that physical
capital was identified early on as a growth factor as well as the
fact that it's fairly easy to measure and, therefore, to study.)
And in looking at physical capital, economists have traditionally
argued that the main reason some nations have more of it than others
is their very different savings levels. "The central problem
in the theory of development," wrote W. Arthur Lewis, a Nobel
laureate considered the pioneer of development economics, "is
to understand the process by which a community which was previously
saving 4 or 5 percent of its national income or less, converts itself
into an economy where voluntary saving is running at about 12 to 15
percent of national income or more."
Without adequate savings, nations have little to invest in physical
capital and so little ability to develop. For over 50 years, therefore,
economists have analyzed impediments to saving, including high dependency
rates (high ratios of young and old people to those of working age)
or a fundamental inability to produce more than subsistence levels
of food and shelter.
Other economists have looked more directly at investment per se, rather
than savings. Chari, Kehoe and McGrattan, in a 1997 Minneapolis Fed
staff report ("The Poverty
of Nations: A Quantitative Investigation"), for example,
argue that disincentives to investment arising from corruption or
bureaucratic inefficiency, say, represent distortions in investment-output
ratios. "Distortions of these kinds typically raise the prices
of investment goods relative to those of consumption goods,"
they wrote. "Indeed, in the data, investment goods prices are
low in rich countries and high in poor countries."
Klenow, working with Chang-Tai Hsieh of Princeton University, takes
a closer look at the data in a May 2003 National Bureau of Economic
Research working paper, "Relative
Prices and Relative Prosperity."
When economists examine the relationship between income levels and
investment rates across countries, note Hsieh and Klenow, they typically
value the quantities bought in each country at "international"
prices—that is, at the average prices prevailing across countries.
That choice is critical.
Economists use the concept of purchasing-power parity (PPP) to arrive
at international prices. The idea is that exchange rates don't fully
reflect the differences in standards of living among nations, so taking
the price of a commodity in one country and doing a simple exchange
rate translation into another country's currency doesn't necessarily
give you a good sense of the cost of that commodity in that other
country. A tuna sandwich in Tokyo costs more than it does in Topeka,
simply because the cost of living is higher in Japan, regardless of
the exchange rate (or the relative scarcity of tuna, for that matter).
PPPs are meant to deal with these differences, attempting to provide
a proper basis for comparing living standards among nations.
"A PPP is the rate of currency conversion that equalizes purchasing
power of different currencies," according to the Bureau of Labor
Statistics. "[It] has the dimensions of an exchange rate as
well as a price index." In effect, PPPs are like a cost-of-living
index across nations—applying the same measuring stick to all
countries by adjusting for standards of living—just as the consumer
price index allows economists to see how price levels change over
time for the same shopping basket of goods and services.
While PPP prices are essential for some purposes, they're misleading
for others. "If you want to compare purchasing power of incomes
in different countries, PPP prices are helpful," said Klenow.
"But they're not necessarily a good guide to why investment
rates differ across countries." Understanding investment rate
variation among nations calls for an analysis of domestic
prices, not international prices, argue Hsieh and Klenow, because
most people base their purchasing decisions on the domestic prices
they face in the home market. (Domestic and international prices might
be the same for some products, but not all.)
In making their argument, Hsieh and Klenow first confirm that there
is a close association between investment rates and per capita income
when measured in PPP prices. (For statistics buffs: the Pearson
correlation coefficient is 0.50.) When measured this way, wealthy
countries show an investment rate two to three times higher than poor
But the correlation virtually disappears when looking at investment
rates at domestic prices. (The correlation coefficient drops
to 0.05.) "When evaluated at domestic prices, richer countries
have only modestly higher investment rates than poorer countries do,"
they wrote. "At domestic prices, poor countries like Mali and
Kenya do not invest much less than rich countries such as the U.S.
This, they point out, contradicts the long-held notion that poor countries
are poor because of inadequate savings compared to rich countries,
so all the hypotheses behind those alleged paltry savings rates lose
their importance. "This evidence undermines explanations involving
discount rates, subsistence consumption, low-savings traps, and the
taxation of capital income," they wrote.
The problem, argue the economists, lies instead in the much higher
price of investment relative to consumption (at domestic prices) in
poor countries compared to rich. So saving and investing a given fraction
of local income does not buy as much equipment for businesspeople
in Kenya as in Norway.
And that leads to their second surprising discovery: The reason for
the high relative price of investment in poor countries isn't that
investment goods are notably more expensive in poor nations compared
to rich countries, according to the data. "Computers in Mali
cost roughly the same as in the United States," noted Hsieh
The explanation is that consumption goods—haircuts and taxi
cabs, for instance—are far less expensive in poor nations than
in rich countries. "What is responsible for the high relative
price of capital in Mali is that nontradable services are much cheaper
in Mali than in the U.S." And this finding undercuts theories
of underdevelopment related to investment distortions. The problem,
argue Hsieh and Klenow, isn't that investment goods prices are deformed
by tariffs, taxes or corruption, but rather that "consumption
is cheap in poor countries, making investment expensive and lowering
PPP investment rates."
"Our results imply that the covariation of physical capital
investment rates and income arises from a deeper productivity puzzle,"
they concluded. "The challenge is not just to explain lower
overall productivity in poor countries, but to explain low productivity
in investment goods production relative to consumption goods production."
If Hsieh and Klenow have thus cast a large shadow over conventional
explanations for international differences in development and begun
to shed light on other possibilities, it's apparent that their contribution
is due in part to a willingness to look beneath the macroeconomic
surface. For five decades at least, economists have taken for granted
the idea that investment and savings rates correlate directly with
national income levels. Klenow wasn't content. "The data sets
out there are really easy to use if you don't ask a lot of questions,"
he said. "Looking into the details made us see things that we
hadn't really seen people talk about before."
Klenow brings similar curiosity and technique to bear on a pivotal
question regarding the effectiveness of monetary policy: Is money
Two and a half centuries ago, British philosopher David Hume argued
that increases in money supply affect economic output.
"[I]n every kingdom into which money begins to flow in greater
abundance than formerly, everything takes a new face: labour and
industry gain life; the merchant becomes more enterprising; the
manufacturer more diligent and skilful [sic]. ..."
It was the first formal statement of what economists would later
call the non-neutrality of money, meaning that changes in a country's
quantity of money will also affect the quantity of various goods
and services produced by a nation, at least in the near term.
The theory is based on the idea that putting more money into an
economy will prompt people to spend more. Increasing demand will
induce suppliers to hire more employees and provide more goods.
So in the short run, an increase in money supply will result in
an increase in output and employment.
This can only happen, though, if prices don't jump immediately with
the increase in demand. And that might be possible because of "sticky
prices"—the theory that prices of goods and services
don't adjust instantly. Wages might be sticky, for example, because
of labor contracts. And business owners might want to avoid the
expense of reprinting restaurant menus or product catalogs, so other
prices will be slow to adjust.
Arguing that prices adjust more slowly than output after a money
supply increase, Hume wrote, "It is easy to trace the money
in its progress through the whole commonwealth; where we shall find,
that it must first quicken the diligence of every individual, before
it encrease [sic] the price of labour."
But many economists have argued that Hume (and his modern supporters)
was wrong: Money is neutral. According to this school of thought,
prices are actually quite flexible, and so both demand and supply
curves shift quickly in response to increases in money supply. A
new equilibrium will be established—with higher prices, but
the same quantities demanded and produced. Output and employment
won't change when money supply grows or shrinks (at least not when
those money supply changes are expected).
These are the arguments of the rational expectations school, led
by Robert Lucas, the University of Chicago economist who has powerfully
challenged the concept of sticky prices and the alleged non-neutrality
of money since the 1970s. In his 1995 Nobel award lecture, "Monetary
Neutrality," Lucas observed that new facts had been revealed
through the application of mathematical rigor, statistical power
and dynamic equilibrium models—techniques that were unknown
in Hume's day.
"I think the fact is that this is just too difficult a problem
for an economist equipped with only verbal methods," Lucas
noted, "even someone of Hume's remarkable powers." Aided
by more powerful methodology, Lucas and others have made a strong
case that changes in monetary policy (if anticipated) can
change prices, but not output or employment, because on the whole,
people act rationally, markets perform efficiently, and prices adjust
quickly. "Anticipated monetary expansions" have inflation
effects, argued Lucas, "but they are not associated with the
kind of stimulus to employment and production that Hume described."
(Unanticipated monetary changes, on the other hand, could
have short-terms effects. Still, said Lucas, "none of the
specific models ... can now be viewed as a satisfactory theory of
Economic journals are filled with debate over monetary neutrality,
but the argument is far more than theoretical. When the Federal
Reserve's Open Market Committee discusses monetary policy, its decisions
about discount rates and federal funds rates are predicated on assumptions
about the likely impact of its actions. If the FOMC lowers the federal
funds rate, leading to an increase in the money supply—or
if it does the opposite—will it affect output and employment,
or just prices?
Klenow engages the debate by testing whether the standard macroeconomic
model used by sticky-price proponents accurately predicts the effect
of monetary policy changes. And to do so, he exploits the fact that
products differ in degree of stickiness—gasoline prices change
quickly, for example, while prices of books and pharmaceuticals
In a winter 2003 Quarterly Review article with Mark Bils
at the University of Rochester and Oleksiy Kryvtsov at the University
of Minnesota, Klenow takes a standard sticky-price model, alters
it to accommodate multiple consumer goods with varying levels of
stickiness and then looks at predictions generated when this model
economy experiences a 1 percent increase in money supply ("Sticky
Prices and Monetary Policy Shocks").
The model predicts that roughly two months after the money supply
increases, prices of flexible-price goods will increase over 1 percent
while prices of sticky-price goods will rise just 0.2 percent. Quantities
of flexible-price goods consumed will decline by 0.1 percent, according
to the model, while consumption of sticky-price goods will increase
about 0.7 percent.
The quantity effect diminishes over time. The model projects that
a little more than a year after the 1 percent money supply increase,
sticky-price goods will have joined their flexible-price cousins
in experiencing about a 1 percent price rise. And quantities consumed
of both types of goods will, after this 13-month lag, reach the
same levels as before. In other words, the money supply increase
will affect prices and quantities of flexible-price goods more quickly
and dramatically than those of sticky-price goods, and over time,
prices for both will rise 1 percent and quantities will be unchanged.
The reality test
With these model predictions in hand, the economists then look at
reality: What do the data say about price and quantity changes for
flexible- and sticky-price goods following a money supply change?
Sticky-price theory suggests (and the model predicted) that prices
of flexible-price goods should rise initially, more so than for goods
with more rigid prices.
Bils, Klenow and Kryvtsov look at price and quantity behavior for
goods and services that represent about two-thirds of overall consumer
spending (housing, atypically sticky in both price and quantity,
is intentionally excluded from their calculations). Their data, from
the U.S. Bureau of Economic Analysis, span the years from 1959 to
2000. The purpose of the exercise, of course, is to see how these
data appear to react to changes in monetary policy, so the economists
try out several different measures of monetary policy: the federal
funds rate, nonborrowed reserves and the ratio of nonborrowed to total
The first measure, though, is enough to tell their story. Using a
1 percentage point drop in the fed funds rate—the same expansionary
monetary policy step used for their sticky-price model projections—they
find that prices for flexible-price goods decline by almost
1 percent after two months—the opposite direction from the model's
prediction. "Prices for flexible goods actually decrease relative
to prices for sticky goods in the first eight months after the [fed
funds rate decrease]," noted the economists. "We see no
impact ... on the relative quantity consumed for the first year. ...
Neither the initial nor subsequent responses match those predicted
by the model."
The findings are "robust," as economists like to say:
Altering a few details of the experiment doesn't change the basic
results. Looking at prices from just the last 18 years rather than
the four-decade data set, for instance, changes little. And the findings
are consistent when they examine responses to the other measures of
monetary policy change. Regardless of the monetary policy variable
measured, or the time span examined, a change in money supply appears
to have an anomalous impact on the prices and quantities of goods—sticky-price
goods and flexible-price goods seem to move in ways they shouldn't,
according to standard sticky-price theory.
It's a striking discovery, but the implications are ambiguous. Perhaps
the sticky-price model they've used—standard issue for monetary
economists—doesn't accurately describe how price stickiness
works in response to monetary policy changes. Or maybe the measures
they've used to represent changes in monetary policy—again,
standard issue—are inadequate. And of course, it could be that
both interpretations are valid. "Put more succinctly,"
they concluded, "we reject the joint hypothesis of sticky-price
models and these popular means of identifying monetary policy"
Does this mean that Hume was, indeed, wrong? Is money neutral, after
all? Klenow is careful to emphasize that rejecting a given theory
doesn't prove its converse. More to the point, Klenow's work doesn't
repudiate sticky-price theory. It simply demonstrates that the model
and measures most commonly used to represent and test the theory need
work. "We're actually not taking a stand on whether or not money
is neutral," said Klenow. "We're not bringing anything
to the table in terms of direct evidence or logic on that question.
... Our view is simply that we need to do more theorizing about how
money can affect output."
The bigger picture
In his graduate work at Stanford University, a professorship at
the University of Chicago's Graduate School of Business and the last
three years at the Minneapolis Fed, Klenow has written papers on the
sources of industrial innovation, the relationship between education
and economic growth, and the impact of trade liberalization on product
varieties, among other topics. The central theme, he points out, is
the importance of technology and productivity, especially for economic
growth and development.
His work on sticky prices is thematically distinct, but it illustrates
Klenow's underlying creative process and methodology. "Sticky
Prices and Monetary Policy Shocks" developed out of an earlier
piece of Bils-Klenow research into product quality changes over time,
and how quality is or isn't reflected in prices. "So I wanted
to know exactly how they calculate prices for the consumer price index,
to what extent they adjust for quality and how," said Klenow.
He spent uncounted hours pursuing that exact truth at the Bureau of
Labor Statistics in Washington, D.C.
"The end product," he said (neglecting to mention his
highly regarded American Economic Review paper on quality
and prices): "I'm now a supplemental employee of the Bureau
of Labor Statistics," holding aloft his coveted BLS identity
badge. "So I get to access all these internal documents, get
to find out exactly what they do, and in excruciating detail."
That level of detail is essential to Klenow, because he's unwilling
to accept data at face value. "I happen to be a macroeconomist.
That's my audience and motivation," he noted. "But my
dad and three brothers are accountants, so I guess I have that kind
of background: Exactly how are these numbers put together? Once I
start pushing in that direction I've been able to see some of the
details of measurement that shed light on the macro questions."
(California-raised Klenow has just moved closer to his accounting
roots, leaving the Minneapolis Fed to join the highly ranked economics
department at Stanford University.)
Bringing the data to bear
The hallmark of Klenow's work, then, is not a narrow focus on a
particular area of economics, but rather an intense curiosity about
data and the deep lessons they can tell about the broader economy.
He lays the substance of econometrics over the framework of economic
theory—sometimes modifying theory in the process—and thereby
strengthens ties between micro and macroeconomics.
"I'm looking at micro data and seeing what it has to say about
macro," observed Klenow. "Our core assumption in economics
is rationality at the individual level—firms try to maximize
profits; households try to maximize happiness. Our theories operate
at that individual level. So our models of the macroeconomy should
be consistent with the micro data we observe."
And the end product of the research is not only a firmer grasp on
the reality of development gaps or mechanisms of monetary policy,
but the opportunity to improve them—to facilitate the stability
of economies and the growth of nations. "I really like that
idea of bringing the data to bear," said Klenow. "To say,
here are different hypotheses, the data tell us something about them.
Then we can use that in the future to better understand what's going
on, and to make better policy."
See also: What's sticky, what's not and why?