I gave a speech on
this topic at the Libertarian Alliance in March. A link to the video recording
of that speech is on the main website. The following essay covers similar ground but is not
identical with the speech. I develop the argument differently here. My hope is
that this text is a better articulation of my views and hopefully a good basis
for further debate.

The topic of this
essay is broadly the method of economics: What phenomena does economics deal
with? How do economists develop and test economic theories? What type of
statements are the “laws of economics”, and what are they useful for?

Such far-reaching and
deep issues cannot be covered adequately in an essay, so the following must
remain a brief outline. The position I take on these questions is that of
Ludwig von Mises, whose views remain controversial to say the least, even among
many who are otherwise sympathetic to his positions (including the regular contributors to this blog: David
Ramsey Steele, David McDonagh, and J.
C. Lester, all of whom I rate highly as writers and libertarian thinkers.).

Brief sketch of Mises’ position

Mises argued that economics
was an a priori science. Economics is
fundamentally different from the natural sciences in terms of subject matter,
method, and the type of statements it can make. The natural sciences are (mainly,
at least) empirical sciences (a
as opposed to a priori).
Economics is not an empirical science. It does not discover the regularities that
constitute its laws through observation (for example, the collection and
interpretation of statistics) or laboratory experiments (difficult in
economics) but through careful logical deduction from certain starting
propositions (axioms), although some observation may be involved in establishing
these starting propositions. Because it is not an empirical science, it does
not make empirical predictions and its key theses are not testable by empirical
means. (If this sounds strange to you, don’t worry. Most economists today
disagree with Mises and practice, or believe they practice, a different kind of
economics, though in my view, they are quite mistaken. A lot of what is
presented as “economics” to the public today does not quite deserve the label
and is often intellectually weak. All this will hopefully become clearer soon.)

So is it useful? Yes,
very much so, in fact, it is indispensible. Economic theory is antecedent to
experience (a priori Latin: “from the
earlier”). It provides a tool for understanding
experience, for making experience intelligible in the first place, in this
case the experience of economic phenomena, which are by definition complex
phenomena. Without the laws of economics, we could not speak intelligently
about observed market phenomena, make sense of what happens around us in terms
of economic behaviour, and in any reasonable way collect, organize and
interpret economic data. Economics provides the essential abstract mental tools
(which are necessarily independent of time and space, and thus generally valid)
for approaching a specific situation in real life and dealing with a specific
real-life problem of economics.

The laws of economics provide
a searchlight, a type of X-ray, that illuminates the bony structure underlying all
economic phenomena, the patterns and regularities that are at work in all human
action as it relates to economic goods. (Mises famously defined economics as a
subset of a wider science of human action in general, which he called praxeology,
but for the purpose of this essay I stick to the narrower field of economics.)

“Falsifiability” through experience

As economics so
defined is general, abstract and prior to experience, it cannot be falsified by
experience, which means it is not “testable” or “refutable” in the way that
most natural sciences are, and which has become in fact the generally accepted
definition of what makes an inquiry a scientific inquiry in the first place,
namely for the scientist to come up with testable hypotheses that can be
falsified through experience.

It is this aspect that
most riles many economists (the few that care about the epistemology of their
science), other social scientists, and many epistemologists and philosophers. If the economist produces
stuff that cannot be falsified empirically, so this criticism goes, then he may
either produce tautologies (he simply re-arranges his starting axioms) or
arbitrary nonsense, maybe both, and he can go on repeating it because the alleged
non-falsifiability of it immunizes it against refutation.

In the following I
will try and defend Mises. I believe that his position on the method of
economics is correct. This topic is also extremely relevant, in my view, for any
discussion of economic phenomena and ultimately for any policy discussions. We
need to understand what economics can do and not do, and how it can go about
its business reasonably and intelligently.

Importance and examples

It is important to
stress from the start that Mises did not suggest that this method should be adopted, or that this was a
method that should distinguish the Austrian School economists from other
economists; that this was one available method next to others, and that the a priori one was just better than any
alternatives. He claimed – correctly, in my view- that this was the method of economics. All the key
tenets of economics developed over 300 years of systematic economic investigation,
from Cantillon to Hume to Adam Smith to Ricardo to Carl Menger, were of such an
a priori nature. Observation might
have led these economists to develop their theories. Observation might have
provided an initial spark; provided ideas or aroused interest. But economic
theory proper is always logically derived from human conduct; it tells us
something about the logic of action. To be an economist in the sense of being
an economic theoretician is to think in terms of a priori concepts. To apply the laws of economics to specific
economic problems in a given situation is to apply again a priori concepts. Some examples may help illustrate this:

The law of diminishing
returns is an a priori law. It is
essential for any analysis of real life economic situations. If economics were
an empirical science and if its laws were subject to empirical testing, must we
not fear that tomorrow we might encounter a maverick economic good to which the
law of diminishing returns did not apply? This is impossible. An economic good
that was not subject to the laws of diminishing returns would not be an
economic good, as there would be no reason to economize on it. Being an
economic good means being subject to the laws of diminishing returns.

Interest is the
phenomenon that we value the same or similar goods differently if they are
available at different points in time, and specifically, that we value future
goods lower than present goods. If economics were an empirical science, would
we not have to fear that tomorrow we encounter a group of people to whom that
law did not apply? No, this is impossible (at least as long as these people
have not yet discovered the secret of eternal life). Interest follows directly
from time preference, and time preference is an integral part of what
constitutes an economic good. “To want something means to want it, all else
being equal, sooner rather than later” (George Reisman). Or, to put it
differently, to be indifferent as to whether you enjoy a good today, in five
years time, or in twenty years time, is equivalent to not caring about it,
which means, the good in question is not an economic good to you in the first
place. “Economic good” means you care about it, which means you experience time
preference in relation to this good, which means the concept of interest
applies to it, which means all a priori laws
of interest apply to it.

Ricardo’s law of
comparative cost has led to the law of association, which shows that it is
advantageous (always marginally wealth-enhancing) for individuals and groups of
individuals (nations) to engage in cooperation via free trade, even if one or
more members of this trade network do not possess any comparative advantage,
meaning their marginal productivity is lower in every relevant area than the
marginal productivity of other members. Even then, everybody benefits from the
inclusion of this member/these members in the network. This law is ultimately
derived from the law of marginal utility, of which the law of diminishing
returns is a subset. If economics were an empirical science, would we not have
to fear that tomorrow we might encounter a group of people to whom the law of
association did not apply? That is, again, impossible. Wherever there is trade
in economic goods, to which the laws of marginal utility apply (otherwise they
wouldn’t be economic goods), the law of association applies. The law of
association is a priori. It is not
subject to empirical testing but is logically derived from the key tenets of
economic action and it is therefore a tool for making something as complex as
trade and cooperation of people with different skills intelligible in the first
place. (It is also the most powerful argument against any type of restriction
to free trade.)

Let’s take an example
from the field of money. David Ramsey Steele who is very knowledgeable about
Mises and generally sympathetic to his work but rejects Mises’ apriori-ism, correctly
emphasizes in his book and also recently in his blog entry on this website, that it is often difficult to ascertain
precisely which financial assets fulfil the function of money at a specific
time and place. Over time, what is used as money has changed, and in a modern
economy with a highly developed financial system the distinction between money
and non-money can be blurred. Seemingly safe assets for which very liquid
markets exist have occasionally assumed the role of quasi-monies. However,
these are not problems of economic theory but of application of theory to
specific situations. Once we ascertain what is used as money in a specific
economy at a specific time, the laws of money as specified my monetary theory
necessarily apply to this form of money.

Money is a medium of
exchange, a facilitator of trade that is so widely used by the public that
money prices also function as a reasonable basis for economic calculation and
the monetary asset itself frequently as a store of value. Because money is the
medium of exchange, demand for money (cash balances) is demand for readily
exercisable purchasing power, and any changes in money demand can therefore in
principle be met by changes in the purchasing power of the monetary unit and
therefore without any additional production of money. At different purchasing
powers the same quantity of money can satisfy different degrees of money
demand, something that non-money goods cannot do. If economics were an
empirical science, would we not have to fear that tomorrow we might encounter a
form of money that was a maverick and that would falsify this rule, or to which
this rule would not apply? No, this is nonsensical. By determining that
something is money we also imply that changes in demand for it can be met by
changes in its price. This is a priori.

Are these not simple
tautologies? Are we not re-stating what is already entailed in the original
concepts? – In a way yes. Mises was quite ready to accept this. We may label
them tautologies but it does not make these deductions trivial, useless or
arbitrary. In fact, more elaborate theories can and should be built on these
basic theories. These theories are the keys to unlocking the logical structure
inherent in all economic activity. As we are all economic actors we experience
and we use these concepts daily, mostly without much further reflection. But the
economist illuminates the underlying regularities and laws of economic action and
makes intelligible the patterns that are necessarily at work.

Mises was not
proposing a new method of economics. He was clarifying what it meant to do
economics. As economics is a fairly young science (Mises said the youngest of
all the sciences), it does not have a long history of epistemological analysis.
Many of those who thought carefully about how economics works as a science came
indeed to conclusions that are similar to Mises’: Nassau William Senior, John
Stuart Mill, John Elliott Cairnes (maybe closest to Mises), Frank Knight, and
Friedrich von Wieser. But Mises argued this position most consistently and

What about modern mainstream economists?

Most modern economists
appear to be doing something completely different. They are evidently using
copious amounts of statistical data and applying mathematical procedures to it.
How does this relate to what we just said?

Statistical data
always describes historical events. It shows us what happened in a specific
place at a specific time. It is impossible to approach statistical data, and to
organize and interpret it without any theory whatsoever. Theory-free
statistical analysis is impossible. The combination of (correct) a priori theory and historical data may
allow us to understand what happened in the past, including the very recent
past. Following Steele (as referenced before) we may, for example, try and
understand to what degree liquid AAA-rated floaters functioned as near-monies
in the run-up to the 2007 financial crisis. Or we may analyze how consumer good
prices and producer goods prices behaved in the United States from 1929 to
1933. Or we may try to quantify the extent to which QE has probably lowered the
borrowing costs of the state over the past 5 years.

What we cannot do is
two things: We can neither verify nor falsify the a priori laws of economics, such as the ones listed above. Even
more important (and potentially disappointing to those who derive their
expectations as to what science is all about from the natural sciences) we
cannot derive the laws of economics from mere observation, and that includes
even the most extensive collection of data and the most elaborate and
sophisticated analysis of it. The economists who claim to do this are either
confused or simply play to the gallery (Piketty?) and are frequently not really
proper economists, although some of them may even win Nobel Prizes. This may
sound harsh but I believe it is true. The reasons for why we must fail to
achieve these two things (test/verify/falsify economic laws and discover
economic laws through statistics) are fundamental and I will give them below.
Of course, if economics were a natural science, if it were an empirical
science, these two things would not only be possible, they would be essential
to its modus operandi as a science. Crucially, economics is not an empirical
science in the sense that the natural sciences are.

This is in fact the
reason why no amount of data mining and statistical analysis will ever settle
disputes in the field of economics. Keynesian economists will forever quote
historical data from around the Great Depression as evidence of their crisis
theories and policy recommendations, just as those who subscribe to monetary
explanations of the business cycle (as we “Austrians” do) will forever cite the
same or similar data in support of their theories. It is a common complaint
that anything can be proven with statistics, and in the field of economic
debate this seems to be true to a large degree. (I subscribe to the “Austrian”
explanation of economic crises not because it fits the data better but because
it fits the principles of economics, the laws of economics that allow us to
analyse the cycle in the first place. A detailed analysis of Keynesian theories
leads to conflicts and mismatches with some key economic principles. This makes
this theory much less convincing.)

Paul Krugman, praxeologist?

Any serious discussion
of economic matters must ultimately drill down to first principles, to the a priori level. This means that every
economist is ultimately forced to use Mises’ method, even someone like the
belligerent archtypical Keynesian Paul Krugman, who would not want to be
associated with any tenets of the Austrian School. But if we discussed an
economic issue with Paul Krugman long and hard enough, and assuming for a
moment the honest intention on both sides to get to the heart of the matter, we
would ultimately have to arrive at the level of fundamental theory.

In the 1990s, Paul
Krugman was known as a free-trade Keynesian. When financier Sir James Goldsmith
published his anti-free trade pamphlet “The Trap” in 1994, Krugman attacked it
and Krugman correctly pointed out that Sir James failed to grasp even the basics
of trade. Appropriately, Krugman referred Goldsmith to Ricardo’s work and the
great economists’ essential a priori insights
as to the benefits of trade, benefits that must even accrue to allegedly
“inferior” (less productive) trading partners (see my earlier point on
Ricardo’s theorem). Ricardo, who had then already been dead for 170 years, did
not have the better data but the better theory, as Krugman rightly

Because of the very
nature of its subject matter – purposeful human action as opposed to natural
phenomena – economics can, on the one hand, make incredibly powerful generally
valid statements about human action of the kind of “no country can lastingly be
a loser in free trade”, or “no minimum wage law can lastingly improve the
material position of those on lower income”, which are true regardless of time
and space, while, on the other hand, it cannot make the type of statements, in
particular specific predictions, that one is used to from the natural sciences,
such as “if the price of chocolate goes up by X, demand for chocolate will go
down by Y”, or “if the minimum wage is raised by a unemployment will go up by b”,
or “if the central bank doubles the money supply, the prices of milk will go up
by Z, and the average wage by M.” (To solve these problems, we usually use
entrepreneurs and speculators, not economists.)

Those who are
disappointed by the latter and do therefore not consider economics
“scientific”, unfortunately close their minds to the deep insights that can be
derived from proper economic theory correctly applied.

Economic science versus natural science: The
fundamental difference

“’All daffodils I have
seen have been yellow, so the ones I have still to see will probably also be
yellow’; refinements apart, the generalizations of natural science all rest on
reasoning of this type, and none of them are certain, in the sense that we can
see them to be necessarily true.” (Brand Blanshard, Reason and Analysis, 1962).

Natural scientists
observe that A always coincides with B and make inferences from this
“coincidence”. In analysing inanimate objects and instinct-driven non-human
animals, this has been a very powerful technique. Why? – Because in the
“natural world” there appear to be many regularities and reasonably stable
relationships that allow us to make these inferences. Or, to put it
differently, the natural world does not know valuing, purposeful behaviour, or
“free will”. This changes fundamentally when we introduce human action.

Humans appear to be
unique in that they consciously act, that is, evaluate a situation, make
choices, purposefully interfere with their surroundings, and consciously
re-shape part of their environment. At the core of this process is the act of
valuation, of preferring one thing to another. None of this is observable in
non-human affairs. It is the unique feature of human action, and human action
itself (not the consequences of it in the physical world) is the subject matter
of economics. As Mises pointed out, one day we may be able to determine which
chemical or physical processes cause a person to prefer A to B in a specific situation,
but until we have done so there remains an unbridgeable gap between natural
phenomena and the phenomena of human action, and they require fundamentally
different techniques (this is called methodological dualism). When dealing with
humans we have to assume an element of “free will”.

Put in the same or
similar situation, two people may value and act differently, and the same
person may value and act differently at different times. (This does, at this
stage, not even relate to David Ramsey Steele’s points about rationality or
consistency of preferences, as raised in his blog.) Furthermore, the complexity of the
world we experience as humans means that we can certainly not “step into the
same river twice.” If people responded to the rise in chocolate prices one way
in the past, they may still respond differently the next time. But all that
observation of history (statistics) can do, is ascertain how they acted at a
specific time and place. The problem is simply is that people are not
automatons, billiard balls, light rays, or amoebae. They do not respond simply
to stimuli.

This puts the student
of human action at a disadvantage to the natural scientist in one respect – namely,
that he or she cannot assume the stability of observable relationships in the
same way that the natural scientist can – but also provides a fundamental
advantage: The scientist is himself or herself a rational human being, and as a
social scientist uses human reason to analyse rational human behaviour. While
the natural scientist remains forever outside the very forces he observes (he
or she never has access to those “prime movers” that make billiard balls behave
the way they do), the economist (or praxeologist) can relate to what he or she
observes in a much deeper way.

It is, I hope, now
becoming clearer, why the laws of economics are of the nature they are, that
is, a priori, as illustrated before. They
reflect the inherent logic of rational behaviour and are thus essentially
restatements and careful further elaborations of the starting axioms that man
prefers one thing to another, that he values and then acts. Notions such as
economic good (and therefore marginal utility), time preference (and therefore
interest), cost, benefit, profit, loss, are all logically deducted from these

The a
in natural sciences

When one raises the
issue of the a priori in economics there
is usually a lot of pushback from natural scientists and this seems to reflect
the harsh treatment the a priori concept had to endure in their discipline in the
20th century at the hand of the philosophy of analysis, of logical
positivism and extreme empiricism. A
concepts had always been indispensible for making sense of things,
including natural phenomena, but particularly since David Hume there has also
been doubt as to the validity of these concepts and their ability to tell us
anything meaningful about reality. This scepticism was taken to new extremes in
the 20th century. The question was raised whether the standard tools
of abstract human reasoning, such as logic, mathematics, and geometry, that is,
the classic a priori disciplines, did
even meaningfully correspond to anything in the real world at all. Bertrand
Russell seemed to have had this in mind when he said: “I thought of mathematics
with reverence and suffered when Wittgenstein led me to regard it as nothing
but tautologies.” Or, as Einstein said: “As far as the theorems of mathematics
refer to reality, they are not certain, and as far as they are certain, they do
not refer to reality.”

A key event behind
this new trend seems to have been the discovery (by Bolyai and Lobachevsky) of
non-Euclidean geometry, which seemingly knocked the undisputed paragon of a priori thinking of its pedestal. Here
was the former superstar of a priori reasoning:
entirely man-made, a creation of abstract thinking, yet a powerful and
indispensible tool for dealing with the real world: Euclidian geometry. But
after a more than 2,000-year unassailable reign it now had to face a newcomer.
Today it seems to be widely agreed that Euclidian geometry is still useful for
building bridges on earth but that when it comes to analysing big stuff in
space, the new type is much better.

Should this shake our
faith in the a priori method in
economics? – No, said Mises. First of all, the idea that the tools of human
reasoning are often just adequate rather than perfect did not surprise or shock
Mises. “There is no such thing as perfection in human knowledge, nor for that
matter in any other human achievement. Omniscience is denied to man. The most
elaborate theory that seems to satisfy completely our thirst for knowledge may
one day be amended or supplanted by a new theory.” (Human Action, Introduction)
But importantly, the arrival of a new geometry did not mean that these a priori concepts were just arbitrary or
simply the result of convention. We could not redesign a new geometry at will.
In fact, to Mises, the history of geometry showed that the human mind was quite
capable of developing a priori concepts
that did meaningfully respond to the real world (as Euclidian geometry still
does, even now that it has lost some of its lustre).

And furthermore, the
epistemological problem of whether or to what degree the products of abstract
human reasoning correspond to the physical universe may trouble the natural
scientist, but it is in fact irrelevant for the economist. As we have seen, the
economist deals with purposeful human behaviour, with human rationality at work
(and, again, not even with the physical manifestations of human action, but with
the inner logic of human action). He or she applies human reasoning to the work
of human reasoning. In the “natural world” there may be no self-evident truths
that allow for any necessarily valid deductions. Here, the human mind must
always guess. But, to the human mind, rational human action is the ultimate
self-evident truth, and here necessarily valid deductions are not only
possible, such a priori inferences
are the only statements we can make with any certainty.

There is, of course,
much more to say about this important topic but I hope that the above provides
at least a good basis for further discussion. The last word belongs to Mises:

“…the sciences of
human action differ radically from the natural sciences. All authors eager to
construct an epistemological system of the sciences of human action according
to the pattern of the natural sciences err lamentably.

The real thing which
is the subject matter of praxeology, human action, stems from the same source
as human reasoning. Action and reason are congeneric and homogeneous; they may
even be called two different aspects of the same thing. That reason has the
power to make clear through pure ratiocination the essential features of action
is a consequence of the fact that action is an offshoot of reason.” (Human
Action, Chapter II).