Arnold Kling  

The Trouble with Macro

The Fiscal Outlook... Cap and Trade: Be Afraid...

My earlier post suggesting a macro-less economics major drew some interesting comments. I agree with those who say that a student should choose courses on the basis of the professor rather than the topic--that was exactly the advice I gave to my daughter.

On the issue of dropping macro, I think it boils down to three questions.

1. Do you think that it undermines otherwise sound economics?

2. How useful and important is AS-AD?

3. How useful and important is IS-LM?

In my view, macro seriously undermines sound economics. It treats work as scarce and consumer wants as insufficient, which is the opposite of what we teach in micro. Macro treats saving as contractionary and international trade deficits as contractionary, which is contrary to general equilibrium micro. Most people with no economics training intuitively believe that jobs are scarce, that they help the economy when they spend rather than save, and that trade deficits are bad. In general it is the job of economists to explain why those views are fallacies.

Many economists believe that the metaphor of aggregate demand and aggregate supply is very useful for explaining output and inflation. I am not so sure.

One nit to pick is that the aggregate demand curve has the price level on the Y-axis, while the aggregate supply curve has the rate of price change on the Y-axis. I won't dwell on that issue.

My problem is that I think it is really off track to suggest that the economy produces a single good, and that employment variation is due to variation in the demand for that single good. Instead, I think of the economy as having many sectors with constant shifts of demand among sectors. Variation in aggregate employment reflects sluggishness in shifting across sectors. I admit that one must strain to tell this story for the Great Depression.

The IS-LM model of money demand, saving, and investment is an attempt to explain interest rates and aggregate demand. Since I already don't like the concept of aggregate demand, I'm not so excited.

For interest rates, there is again a nit. The LM curve has the nominal interest rate on the Y-axis, and the IS curve has the real interest rate on the Y-axis.

I think that one of the most important elements of the capital market is the risk premium. In fact, there are many risk premia. As the risk premium changes, we see variations in the type of investment that is profitable to undertake.

(My story of the Great Depression would be that there were major shifts away from investments in electric power plants, housing construction, and other businesses, because the risk premium rose. Labor never got around to moving toward sectors that were less dependent on a low risk premium.)

The combination of IS-LM and AD-AS has so many degrees of freedom that one may tell just-so stories about any conceivable combination of macroeconomic variables. But most economists do not believe that the predictions from macro-econometric models based on those concepts have been sufficiently accurate to validate the models. To me, that suggests a scientific failure.

Comments and Sharing

COMMENTS (22 to date)
Matt writes:

Macro is an attempt at a two eigenvector approximation, and should come with a bounded error. This assumption creates an academic model that has successive degrees of accuracy as the decomposition takes on more dimensions.

The interest rate yield curve should be treated as the generalized yield curve, and that yield curve has to be related to the spectral characteristics of individual production functions (eigenvalues).

In other words, there is nothing unique about the money industry as opposed to the "height" industry (Mankiw) or the oil industry or government industry. All industries follow a power law production cycle which is discovered once the economist closes the covariance matrix, that is there is a second constraint on cyclical moves. The second constraint explains that phase allignment constantly occurs as sectors time their reconfiguration moves. That is the second constraint result in a computed phase point the competitive agents converge on.

The relationship between economic exchanges and the sector yield curve is constructed as if the economic exchanges had the purpose of diagonalizing the production functions over time, as if economic exchanges were the estimates of an adaptive kalman filter that is converging the production function.

That we have no equilibrium points results from a basic theory which says the force of efficient market equilibrium causes economic activity to disappear, an impossibility that proves instability. The economy moves toward equilibrium but never reaches it.

The lack of equilibrium forces the economists to use quantum mechanical models.

liberty writes:

1. Your nits are not that small (or perhaps they are small but vicious).

2. Aggregation is a serious problem because it aggregates away all the major action that takes place in economics. Even if the highly aggregated models are correct on their assumptions, the mechanism and the details are entirely missing, and this is likely to lead to a misunderstanding of the actual causes. Its so abstracted that its impossible to confirm it with empirical data. And its so aggregated that even if it were all true, it would still tell us nothing about what is really going on in the economy or why.

3. The models tend to completely leave out institutions, so you have some hidden assumptions about what causes the initial curves - but because they are hidden, people interpret the results in such a way that they might (for example) make policy which then changes the underlying institutional conditions that led to the result (if the model was indeed correct).

As an extreme example, the original Solow growth model would predict that the main thing leading to economic growth is savings, not private property protection. You can go ahead and make all property communal and then as long as you enact a policy of forced savings you should be able to outgrow all other economies.

Thanks for posting on the subject. Too rare we dare criticize the whole field of macro, despite its major flaws.

John writes:

"The lack of equilibrium forces the economists to use quantum mechanical models."

If this was sarcasm (I don't know you well enough to tell), you're pretty funny.

If it's not sarcasm, then I obviously overpaid for my Economics education that didn't teach me these quantum mechanical models.

mjh writes:

Every time I read an economics blog entry that talks about macro and/or micro, I find myself having to go back and re-lookup what the difference is. As a result, I don't think I really understand the difference. It's just not sinking in.

I've read the following (and more) descriptions:

I understand that micro deals with the decisions of individuals or companies, and that macro aggregates those decisions and talks about the bigger effect. But this doesn't really illuminate much. As AK's post suggests, there appears to be tension between micro & macro. Is macro considered to be roughly equivalent to Keynes?

My apologies if this is a silly question. But is there anything that helps illuminate the differences between micro & macro and the consequences of those differences?

MattS writes:

In my intermediate macro class, the IS/LM and AD/AS stuff was counter balanced by growth theory. If a student was looking for it, he would see that these problems were all short run phenomenon and in the long run they aren't problems at all (according to the models we were studying). This wasn't emphasized by the teacher, however. What was emphasized is that the market will not automatically adjust the savings rate to the golden rule level. So I have a feeling most of my classmates still got the wrong impression.

scott clark writes:



michael gordon writes:


Seeing as how I was the poster who criticized Arnold’s lack of any macro-economic course, I will try to reply to his challenge.

Please keep these caveats in mind though: I have a Ph.D. in both economics and political science from Harvard, but was a political scientist all my life, with a specialty in international relations: security, diplomacy, global political economy, foreign policy-making, and --- because of its importance to relative power (military and otherwise) in IR --- comparative kinds of economic systems, not least the historical English-speaking market-oriented economies, the EU welfare-statist economies, Japan’s state-regulated dual economy (advanced and heavily protected backward sectors), and China’s half-capitalist, state-controlled economy. Later, in the 1990s, I delved deep into economic development generally, not least because of its ignorance (with a few exceptions) of the institutional basis of each economy.

Arnold has written intelligent and informatively on the latter, following Douglass North, but tended only briefly (if I remember correctly) to link the wider concerns of a societal or national culture to institutions. National cultures in turn help explain --- help, not fully account for --- the range of political ideologies in each country: why, for instance, do the Continental EU countries in West Europe have a much larger state-role, sustained by socialist-influence political parties and right-wing versions of statist paternalism (Gaullism, say, in France or Christian Democracy in Germany), whereas these influences never or hardly existed in Britain, let alone the US.



Always useful to begin with a clear definition of a controversy, which (I trust) isn’t seen as tendentious. Macroeconomics is that branch of economics focused on the long-term growth of a national economy and, especially in the short-run, fluctuations in prices, output, and employment. Both lead to questions about the role of the government (state if you want) in both the long- and short-terms.



(2-i.) Nobody disputes that long-term economic growth is largely a matter of an efficient private sector: especially the inputs of capital investment (productively employed), technological change, and (more controversial but sound in my view), vigorous entrepreneurship. The reason for the latter’s need? If we’re talking about radical, structurally changing technological breakthroughs that come generally in clustered waves --- roughly every 50-60 years since the first big wave in the 1760s-1820 period or so (iron, steel, textiles, and new energy systems for the machines that produced these) or since 1975 or so in ICT and the Internet --- it takes new dynamic start-ups to generally bring them successfully to the market-place.


2-ii) Our best guide here is probably Joseph Schumpeter’s work on bold risk-taking entrepreneurship, clustered waves of technology (long-term cycles), and the concept of creative-destruction: you can’t diffuse the spillover benefits of big radical technological changes if you can’t free up capital and skilled labor (including maybe managerial know-how) by letting older, more standardized industries shrink or decline or disappear and direct that capital and skilled labor into the new, more productive sectors.

Standard growth-models of the Solow sort completely ignored these crucial elements, other than capital. And they postulated, wrongly, an end-state as capital accumulation occurred by means of diminishing returns, with somehow --- full external to growth modeling (exogenenous to it) --- technological change intervened. New growth theory at least tried to endogenize technological change, but neglected the clustered-wave ideas of Schumpeter, the crucial role of bold entrepreneurship, and above all the key idea of creative-destruction. Only in the late 90’s did Schumpeter’s work enjoy the revival that it deserves.


2-iii.) But the main problem even with Schumpeter’s work is that the institutional background was ignored until recently (except for North and his followers in economics) or taken for granted. So let’s define institutions --- yes, a macro-level matter, and not just confined to the national economy.

Institutions are systems of general rules embodied in various organizations --- say, legal courts and the police, legislatures, executives, civil and miltiary bureaucracies, corporate businesses, educational systems, and professional associations--- that enforce them formally by means sanctions or rewards and can often alter them in rule-based or prescribed manner. That said, the actual ways in which these organizations operate can only be understood by grasping the social norms --- a cultural matter --- that prevail within each and between them. Some of these social norms are specific to individual organizations: say, the ways in which cops on –the-street are honest or corrupt (or worse). Most of these social norms, though, reflect a particular kind of national culture that evolves historically and profoundly shaped the workings of institutions and the attitudes and behavior of their organizational members.


2-iv.) A national culture means what?

A network of beliefs, values, and operational (social) norms about the world and human nature and the wider society around individuals within a particular nation (or local) society: such as whether you can trust others or not, whether the basis of your community is your family or wider family clan, or tribal-clan, or social class (the aristocracy, say, vs. the peasants vs. the urban workers), or maybe , if a country is lucky, in some meaningful way a national community based on shared citizenship. If a fairly widespread sense of trust exists, then in turn large numbers of people can be counted on to cooperate spontaneously for common ends --- within specific organizations and between them and the community in question. If trust doesn’t exist --- or if mistrust and cynicism do on a large scale --- then pervasive corruption and predatory behavior are likely to prevail among power-holders, and social and political conflicts cannot be easily resolves except by ever greater formal sanctions . . . at the extreme, in say virtually every Arab country, only by despots and their favored tribal-clan and wider crony patron-follower networks.

Another example: assumptions about how to get ahead in life . . . a key cultural matter, with specific behavioral outcomes. Do people rise in income, wealth, and status by means of hard-work, education, and professional accomplishments (including starting a new business) or by means of gaining access to a crony patron-follower network, mutual back-scratching, and the orgy of money-grabbing and power-aggrandizement that ensues?

How do cultures get transmitted from one generation to another?

Simply said, by a socialization process that begins in infancy and early childhood, thanks to the formal and informal (ostensive) learning by the young child in its interactions with its parents, in learning a language, in interacting with siblings and teachers and others . . . reinforced through dynamic stages of personality development that can and are studied in a variety of ways. The same holds for why, say, a sub-group of a society may be socialized into highly dysfunctional beliefs, values, and norms of behavior . . . the case of far too many African-American infants and children being born these days into single-parent mother-headed families who live in bad crime-ridden neighborhoods and with no father-figure around to identify and idealize as an internalized "self-object" for guidance in life.


2-v.) Against this background, what do all the rich industrialized countries of the world have in common --- rich defined, say, as $20-25,000 minimum in per capita income translated into PPP (purchasing power parity): all of West Europe, North America, Australia and New Zealand, Japan, South Korea, Taiwan, and Singapore?

With the exception of Singapore, all are functioning democratic states; all are capitalist, but with considerable variations in the role of the state in economic life; all have independent judiciaries, police subject to professional rules and legal norms, militaries under constitutional authority, a considerable amount of free speech, free assembly, and competitive elections; and all have generally well-educated populations, able to work with the most advanced technologies --- even if most countries acquire the cutting-edge technologies by means of imports from a half-dozen of the richest and usually most scientifically renown: the US, Germany, Japan, France, and Britain, with the others showing ability not just to acquire technologies by means of licensing or multinational implants, but able to diffuse them within and across industries fairly quickly. And generally, all protect private property through the legal system; restrict corruption (with fairly big variations, say, between Scandinavia and Italy or Greece); and operate with a sense of shared national citizenship.

Singapore stands out as an exception in democracy and fully free expression and association. On the other hand, it is small and homogeneously Chinese and does very well in importing and diffusing technologies near the technological frontier.


2-vi.) No country in the world that fails to meet these criteria --- certainly not China (a per capita income of under $6,000 vs. the US’s $46,000 in 2007, adjusted for PPP) --- is ever likely to close the gap noticeably without major changes.

In the Chinese case ---- where its real GDP growth rate has always been exaggerated (owing to a combination of dubious inflation-deflators and locally gathered statistics that even the Prime Minister openly questioned in the late 1990s) --- the advance of its economy is in line with standard convergence catch-up growth theory, at any rate as pioneered early in the 20th century and after WWII by Japan and then since the 1960s by the smaller Asian dynamos and possibly India since 1990 or so. Roughly 66-70% of its technologically advanced exports --- including consumer electronics and pharmaceuticals --- are provided by foreign multinationals working in that country, which use disciplined and fairly cheap Chinese labor as an export assembly line.

Then, too, whenever Chinese per capita income looks like soaring , the World Bank discovers --- as it did in 1993 and again two years ago --- 200-300 million very poor Chinese that didn’t appear in the statistics earlier.

A query prompts itself here: Can a country ruled by a cumbersome state, topped by a corrupt and venal CP monopoly --- whose banking system is still overwhelmingly used by the state to funnel the massive savings of the Chinese people into preferred investments (mainly propping up the remaining and ineffectual state-owned industries) --- move within two or three decades into the charmed circle of rich and technologically advanced countries without major political and cultural changes? Meaning? Clear reforms that move the country toward far greater decentralization of economic decision-making to firms and individuals; a rule of law (not least for intellectual property); free expression and the right to discuss, innovate, and move freely in and out of the country; and the self-immolation of the CP monopoly --- when has this ever occurred (not just Communist monopolies, but any power-wielding monopoly in history)?

2-vii.)And so, what can we conclude about long-term growth, a macro-economic matter?

Without explicit studying of these matters --- technological invention and innovation (especially of the radical economy-changing sort), entrepreneurship, waves of creative destruction, and key institutional-cultural maters --- economists who hope to understand why some countries are rich and advanced and others aren’t (and others likely won’t ever become rich and advanced) need to be carefully trained in these matters: both historically and in the present, with constant exposure to not just theory but evidence of all sorts . . . only some of which can be modeled formally and run statistically.


2-viii.) And within the circle of rich industrial democratic countries, which do better and why?
For my part, I add that at my web-site --- --- I posted 7 or 8 very long articles in 2004-2005 on why the less statist economies of the English-speaking world have generally done better in GDP and per capita income growth than their more statist equivalents in Japan or on the Continent of Europe. Britain, Ireland, New Zealand, and Australia all reduced state-spending and especially on welfare while freeing-up more than labor markets, with a noticeable boost in their rates of growth in GDP and per capita income.
None of that work denied that the Continental EU countries couldn’t make their labor-markets more flexible. In fact, since then, the Scandinavians and Germany and France (the latter to an extent) --- have, moving in a commendable increase in the labor-participation ratios of their populations between the ages of 16 and 64. Similarly, they are all trying --- with various degrees of success --- to deal with an ever larger number of retirees in the future (as the potential native labor force shrinks) by expanding the age-limit of full pensioned-retirement . . . almost all of which, on the Continent, is in the form of state-funded pensions.


What they haven’t done, so far --- nor Japan noticeably either --- is expand their rate of productivity. In fact, the most recent study of the increased ratio of labor-force participation in West Europe shows that it has come so far at a reduced rate of productivity. Or (as I posted yesterday at Carpe Diem --- Mark Perry’s laudable web site) in a debate with a vocal big-government advocate of the EU Continental sort:

‘One of the best recent studies of this is by Robert Gordon (an outstanding macro-economist at Northwestern) and Ian Dew-Becker (probably a British economist):

"Europe’s employment growth revived after 1995 while productivity growth slowed: Is it a coincidence?" April 15, 2008 Click here:

Their findings, though, might not make you happy. They note that the improved labor-force participation ratio (and reduced EU unemployment) has come at the expense of productivity growth. It's as though, the note, the EU countries have not experienced yet the "Internet Revolution" that drove up the US rate of productivity markedly starting in the early- and mid-1990s and that has continued ever since:

".... The third and most novel – and also most controversial – finding is that there may have been a substantial trade-off between labour productivity and employment growth over the past two decades. Before 1995, European policy made labour more expensive through higher taxes, tighter regulations, and strong unions. This reduced labour demand, lowering employment but raising the real wage and the average product of labour. Slow employment growth and relatively high productivity growth were negatively correlated. After 1995, this process was reversed, with lower taxes and looser regulations reducing the cost of labour, which helps explain the simultaneous increase of growth in employment per capita and slower growth in labour productivity."

"We find that the revival of European employment growth can help explain why European productivity slowed. But we do not explain why European productivity growth did not accelerate as occurred in the US. US productivity took off after 1995, growing at 0.7 percent faster per year, but in Europe a literal reading of the productivity growth data leads to doubt that the internet revolution ever occurred in Europe. Some of Europe’s poor recent performance can be explained by reforms that will enhance growth in the long run, but not all of it. Our findings should lead EU policymakers to think about the two-edged effects of policy reforms on employment and productivity, but they should also worry about how to encourage innovation and the adoption of new technologies.1"



(3-i.) In the short-term, where fluctuations in GDP, the price level, and employment are the chief concerns, the main questions are whether the government and the central bank (if independent of the state) have harmful or beneficial effects on the fluctuations. If the effects are beneficial --- say, as in the US, recessions (and two Great Depressions in the 1870-1880s-1890s and in the !930s) --- have been reduced in number and severity, and why. If the effects are harmful, as say overzealous stimulatory and regulations of energy were in the late 1970s Carter era were, what did we learn? And did overzealous fiscal stimulation in the Reagan era --- dubbed (right or wrong) supply-side economics --- have the same harmful effects, and why or why not?

(3-ii.) I will pass on the LM-IS model, even though, please note in passing,

I happened to have studied in the early 1960s with John Hicks at Oxford, who invented it. It assumes a fixed, completely horizontal supply curve that seems unreasonable to me, and is inferior to an AD-AS approach

(3-iii.) Let's start with basics.

The Aggregate Demand curve, which appeared (I believe) only in the late 1970s as an alternative to the LM-IS curves model, is intuitively easy to pin down.

At its simplest level, it slopes downward for a trio of reasons:

* Lower prices of goods will entail greater purchasing power of goods and services with a given supply of money.

*Simultaneously, a lower price level will reduce the demand of the public --- consumers and firms --- for money and hence to buy what they normally do. If, then, both consumers and firms reduce their money balances, some or most of the freed money will find its way into interest-earning assets: say, bonds, equities, or savings accounts. In turn, in the loanable funds market where interest rates are determined --- a sounder approach, in my view, than the liquidity-preference model Keynesians like (though the effects on real interest-rate changes should be the same) --- this new inflow of money should lead to lower interest rates for the economy. The outcome? It’s fairly obvious; no need to say more about consumers buying more durable goods on credit or business firms expanding output or maybe buying more machines or new plants. Aggregate output then should expand through this mechanism.

*Quite possibly --- it depends (as everything always does in economics) --- lower prices will lead, say, Americans to switch from buying some imports to cheaper American substitutes. Oppositely, lower American prices should lead in principle to foreigners buying more American exports. Both trends expand aggregate output.


(3-iv.) Enter complications for the AD.

Predicting changes in the short-term AD curve is complicated because it is subject to various changes in the determinant variables, themselves interacting with one another. In the upshot, changes in each are not easy to pin down, and of course that requires a macroeconomist to delve into the mental worlds and hence the expectations and decision-making choices of individual economic actors: consumers, workers, business firms, investors, and political policymakers including even an independent central bank.
What are these various interacting influences or variables.
Five or so seem important enough to single out:

*The real wealth effect (say, families feeling wealthier as they did in the late 1990s as stock prices soared and about 60% of Americans were involved in equity markets).

* Changes in real interest rates.

*Changes in the expectations of both consumers and businesses as to the future of the economy : usually, in the short run, over the next few months or a year.

*Possible changes in the expected rate of inflation, a complication that is made harder to deal with because the public usually focuses not on the CPI, let alone other measures of inflation in the short-term, but on a few chosen goods: say, gasoline and certain (hardly) all food prices today as opposed, say, to clear declines in other food items or consumer electronics (HDTV or pc’s or stereo equipment or cell phones)
*And finally changes in the exchange rates and the imports and exports of foreign economies: meaning foreign business firms and consumers as well as foreign governments.

Their purchases, say, of US exports depend, of course, on their own economic performance and the expectations of their firms, consumers, and government policymaking bureaucracies. (Economic nationalism and specific national tastes may play a role too, though these can change: witness the bursting enthusiasm all over Europe for New World wines: those from California, Australia, New Zealand, and Chile . . . all much cheaper and better tasting than their counterparts in Europe.)

3-v) The Theoretical Controversies: Back To the Expectations and Choices of Individuals Again

To figure out the impact of these 5 variables requires the use of a theoretical analysis of what underlies expectations . . . including expected and unexpected changes in each. If Arnold wants to consider these the core-matters and see them as covered by his list of courses, I’ve no complaint: it’s a matter of linguistic terminology, nothing else. And that includes the study of central banking decisions on the money supply and interest rates (short-term anyway).

Where Arnold and I would disagree, I’m sure, is that he likely postulates rational behavior and follows its theoretical linkages, whereas I am much more impressed by the work of social-psychologists on frame-theory ---including economists influenced by them (three of them have won Nobel prizes since 1970, two just a couple of years ago).

In effect, to grasp how investors, consumers, savers, and others operate in an economy --- or for that matter, state-leaders in dealing with threats from other states --- you cannot just postulate rational-behavior, cost-benefit analysis, or strategic interaction (game theory) as explanatory models. You have to look at how individuals “frame” --- understand cognitively and emotionally --- a particular situation: how they define the status quo as good or bad, how they define risks in dealing with it, why they see losses from the status quo as more serious than why they value the benefits of a similar magnitude of gains, and so on.


Behavioral economics is promising here. Even so, it is macro-economics that alerts us and points up that there may be serious coordination problems across individual consumers and firms and investors and workers and savers (not to forget the central bank and fiscal policymakers) that lead to some sort of massive problem of coordination failures and hence sticky prices in the short-term as well as AD externalities:


3-vi) Sticky Prices, Coordination Failures, and AD Externalities

There are two ways to view the resulting fluctuations in output and employment (as well as inflation ): real-business cycle theory or some problems in the market economy at the micro-level that Mankiw and others emphasize. The controversy continues to flare. No point in delving into it in depth.

* Suffice it to say that real business cycle theory postulates flexible prices and wages in the short-tem, finds that the willingness of workers to take jobs at existing wages depends on their expectations of future wages --- the postulate of inter-temporal labor-substitution --- and argues that the business cycle is a result largely of major technological shocks, with technological regress the immediate cause of recessions: a playing out of older technologies, which leads to reduced business output and incentives to work during the recessionary period.


Enter the further postulate in real business cycle theory that not only does fiscal policy not work to shorten the length or depth of a recession --- a claim made, originally, of course by monetarist economists and now in a new guise of the revived “Ricardian Equivalence” about how increased deficit spending not based on tax-cuts leads forward-looking consumers to save any new income to offset future rises in taxes --- but also the neutrality of money. That’s because the money supply is regarded as endogenous to the economy, and hence the causal links between the money supply and output are from the latter to the former . . . and not, as monetary policymakers like the Federal Reserve think, vice versa.

Here again, no clear evidence seems to have arisen about the direction of causation between the ways in which output and money supply fluctuate.

About all I can say as an outsider is that the recent efforts by the Fed to calm bankers, stock markets, and investors seem to have had a noticeable effect (so far) in restoring confidence in the future direction of the economy. Not all of this has been due to increasing the money supply and reducing short- and so far long-term interest rates; it also involves, contrary to what libertarians would want --- especially those who follow the Austrian economists here rather than Hyman Minsky --- the Fed’s supplying funds directly to certain brokerage houses.


My own view (for what it’s worth)?

It comes down in part to the need for more empirical work, but the empirical work has to be based not on assuming rational-behavior but actually doing survey work of an intensive sort in trying to figure out what really motivates investors, firms, workers, consumers, and monetary and fiscal policymakers: including, of course, how they see the status quo, what they expect about changes in it during the near-future (and why), and how of their choices in the present interact and are coordinated or not.


3-vii) AND SO?

And so, in my view,far better survey work --- which requires economists not to just model aggregate data or not make assumptions about rational behavior (and maybe far-fetched inter-temporal labor substitution) --- is needed, something economists have not been trained historically to do.

Here, moreover, survey work can be supplemented by the kinds of laboratory experiments used for decades now by social psychologists in studying the mental worlds of small groups. These small group experiments lead to a limited number of variables that are easier to control and check than complex regression models.

The latter are subject to all sorts of problems: the data-base may have ambiguous information (what exactly is the rate of real inflation in the US economy and hence real GDP and per capita income growth?) or be subject to debates on the dates used; omitted variables may be at stake, and dummy variables are a very crude way to deal with them; there may be complex interactions among the variables that are doubly hard to deal with in logistic regression (where the dependent outcome variable(s) are qualitative); proxy variables may be fanciful: what exactly is a good proxy? And so on. Yet we then get confident assertions that the model's results are statistically valid at some impressive level (5% or less) or with an error-term so small it takes an electronic microscope to zero in on it.

Note though:

Social psychologists can’t, it’s true, control all their variables. You can never be sure that though experiments can be replicated with different subjects and different experiments that their background experiences might not be at work: e.g., professors using university students as the subjects. But at least these experiments don’t wander off into an abstract wonderland of ever greater statistical complexities.

When is the last time anyone has found a major scholar to specify and test a statistical model that refutes his or her earlier and major theoretical findings?


Well, hell . . . this has taken me about 65 minutes so far --- much more anyway than I had anticipated. Enough anyway has been said for some debate.

Michael Gordon, AKA, the buggy professor:

eric writes:

[Comment removed for supplying false email address. Email the to request restoring this comment. A valid email address is required to post comments on EconLog.--Econlib Ed.]

Snark writes:

If for no other reason than by sheer attrition of commentary, I agree with the buggy prof.

John writes:

I have 7 degrees from Harvard. Well, I kid.
Seriously though, that was an interesting essay (, Michael). I agree that teaching about long-run growth is important, but the emphasis in most of my classes has been either IS-LM, AD-AS models in undergrad or solving representative agent problems with Euler equations in my M.A. courses. We did a little about long-run growth in my intermediate macro course, but its not hard to pick up most of it in Romer's Advanced Macroeconomics. I think you can learn most of the lessons regarding long-run growth without necessarily learning the models and much of it could be lumped in with a class on Institutions.

Generally, I would think that since most econometric models can't do much better than an ARMA model at forecasting, that I wouldn't put that much faith in them. Much of your post discussed points that if they were covered at all in my undergraduate education, they were barely covered with any depth to give the student any real understanding. For example, we learned about the concept of real business cycles, but we didn't learn anything about what those models actually look like or a detailed study of their assumptions and conclusions.

I would say that if you are looking for a solid undergraduate study of Economics and were looking to be an economist, you probably should spend more time taking math classes. Other than that, the most important reason to learn most macro is simply that other people who talk about the economy have all been talk the same things about macro. So take a bunch of math classes and if you're still interested in Macro, read Romer's book to get an idea of what the general opinions are. Best to skip the introductory classes.

Mauricio Flores writes:

i MUST have macro in order to get my undergarduate degree in economics, so well, this really helps, i am having my final exam in a couple of days. two things i have found out about macro: Monetary policy is way too simplistic as it is explained in many textbooks. The AS-AD model looks cool albeit oversimplifying reality, but i gets through the point of neutrality of fiscal and monetary policy.

wish me good luck in my exam! :)

reason writes:

"It treats work as scarce and consumer wants as insufficient"

No it doesn't. It (Keynesian Macro) just says that because of exogenous factors, market imperfections and adjustment delays the circular flow of money in the economy may be out of balance (in either direction).

Are you denying the existance of involuntary unemployment (as against merely denying that the actions commonly taken to counter it are counter productive)? If so then you are a loony. Seriously.

Arnold Kling writes:

I believe that there is involuntary unemployment. I believe that the AS-AD and IS-LM paradigm to explain involuntary unemployment is story-telling, not science. As long as we are telling stories, I would rather tell one about slow adjustment of labor when relative prices change and leave it at that.

reason writes:

Then why are you railing against Macro. Macro just points out certain accounting identities that act as constraints on cumulated micro behaviours. You are really railing about some particular theories in Macro.

fundamentalist writes:

Michael Gordon: "Macroeconomics is that branch of economics focused on the long-term growth of a national economy..."

Not a good definition. As taught Macro assumes equilibrium. That's its major fault. My rejection of mainstream macro came from my study of development, which shows that so much of mainstream macro is nonsense because of its assumption of equilibrium. It also led me to Austrian macro, which every student should learn.

fundamentalist writes:

For a really good critique of IS/LM and AS/AD, check out the George Reisman's book "Capitalism." He has excellent chapters on those concepts.

fundamentalist writes:

Other reasons not to study mainstream macro: it has no concept of time or the capital structure. As a result, it's good for mind games but nothing practical. However, studying macro via Hayek's "Pure Theory of Capital" and Mises's "Human Action" is very enlightening and useful.

Mick Rolland writes:

Aggregate supply and demand (AS-AD) curves have another major problem apart from what Arnold stated: Their essence is a false analogy with supply and demand curves in neoclassical micro. Whereas in the Y-axis in micro curves what there is is in fact the relative price (because we are supposing 'ceteris paribus' that the prices of all other goods are constant), in AS-AD we have the aggregate price level.

If all prices (including wages) rise or go down in unison, there is no reason to think that aggregate demand or supply will change. Then if we want some effects on demand or supply we need to assume that while prices for goods move both wages and money remain constant, which is by all means a heroic assumption. It is scientifically useless and inappropriate for teaching particularly because a general rise in prices is usually best explained by a movement in the supply of money.

But I wouldn't go as far as to suggest that macro is in a worse state than micro. Micro can also be considered in a much less than ideal state since the imperfect/monopolistic competition revolution of the 20s-30s. Most teaching since then draws on the reference of the dubious model of "perfect competition" (a mischaracterization of the thinking of the great Marshall and Knight), combined with monopoly, oligopoly and monopolistic competition models that are either of questionable applicability or directly misleading: the comparison of these models with a "perfect competition" setting is mostly inappropriate (as the 'perfect competition model does not represent at all the most important features of an unhampered market, particularly in how it deals with adjustment, information and risk), and it invariably gives rise to 'market failures' everywhere. Even more questionable is the application of game theory to micro, which can give you any result you want with appropriate tweaking of the parameters (that is, the contrary of robustness).

Most endogenous growth theory and international trade theory with "microfoundations" on weak micro has the same defects.

rjp writes:

What I suggest is learning Macro with Robert Barrow's book. In it, he goes completely through the equilibrium business cycle model.

This teaches much micro and prepares undergrads for grad school (that is, it shows them a model and how one should use modeling.)

Mike Sproul writes:

Macro has its defenders (so does Astrology), but I haven't seen anyone defending our current practice of making macro a CORE course in the economics major. (unless it was buried in one of those hour-long comments above) Macro should be an optional course, if it is taught at all. The resources freed up should be devoted to price theory, econometrics, and economic history.

Peter McCluskey writes:

It's been a bit over 30 years since I experienced how macro was taught, so I'm uncertain whether current macro is harmful. But trying to fix it seems more valuable than trying to eradicate it.
I want to replace the IS-LM model with a model of deflationary depressions as the result of a speculative bubble in currencies. I.e. the dollar is bid up in roughly the same way that stocks are bid up in a stock market bubble. Combined with an endowment effect keeping wages sticky, depressions can be understood in ways that fit well with micro plus behavioral economics. And the relationship between a currency and the rest of the economy is broad enough that the economy with a single good gives a good enough approximation of what's happening.

Macrocompassion writes:

Having examined the article and the various comments, it seems to me that in general macroeconomics is a badly understood subject and that many of the people who are involved with it take different views. This is a shame because the subject plays a vital role in determining governmental policy and consequently it is most essential to get this subject straight and properly and logically organized.

The various terms need formal definition, which I will leave for now, but before going into theory about macroeconomics there are some basic matters that not all macroeconomists seem to recognise. There are the need to take aggregate quantities and to view the "big picture" from sufficiently far away so as to cover it all. The way that this is expressed is by considering the macroeconomics as a system which comprises specific parts that are connected together in a unique way. This process needs one to construct a macroeconomic model and without such a model in pictorial and mathematical form it is almost impossible for serious analyists to communicate. So I am going to mention certain facts about the way that this model is built and about how it can (in my opinion) be used.

The model consists of various sectors or parts (called entities) each of which plays a distinctive independent and characteristic role in the functional behavour of the whole macroeconomic or social system. By starting to build the model in this way one avoids the complication of statistical behavour studies and the doubts concerning their accuracy. If the separate behavours as characterised by their functions are properly represented then the whole system must respond properly and be capable of a true simulation of what actually occurs.

In the past the models that were used have two or possibly 3 entities (which should not be confused with their communication links (markets) often includes as if they were functional elements instead of simple lines for money and goods reciprocal transfers). These models led to the IS/LM method of envisaging the system, however the more complete model which I am considering here has additional entities (actually 6 or 7 depending on whether the rest of the world is to be included in the model for a particular country).

So the current discussion about the use of the IS/LM theory to predict economic trends and behavours is in my opinion unsuitable due to the degree of oversimplification in the associated model.

What I am proposing for the model has the following entities, each of which is needed to play a separate and particular role in the functional performance of the whole system:

Land-Lord, House-Holder/Worker, Producer, Capitalist, Government, Financial Institution and possibly the Rest of the World.

My analysis requires the first 6 entities to be taken with only 19 specific lines of mutual communications between them each of which exchanges mone for goods, services or other valuable documents. The results are in equilibrium in so far as the values of good etc flowing into an entitity must equal the value flowing out from that entity. However the way that these flows are distributed partly depends on the topology of the connections and also on the decision-making by the entities who respond the various changes in what they receive.

You can see that this model is more complex than what Keynesian theory has previously shown, yet it is sufficiently simple for hand-calculations to be used without needing a computer. This model has shown the effect of introducing various exogenous changes to the system and in particular the short-term effect of different kinds of taxation policies. I will not present my finding except to mention that the operation of this model is based on certain rules which take on a logical and somewhat mechanical nature.

I hope this very limited explanation is enough to provide you with a better realization of what macroeconomics is all about and that the more detailed problems associated with companies, monopolies and national investment are covered in the most general manner, without having the wood rendered invisable due to the numbers of these trees.

I find that when such a model is formally presented and understood in basics, then a lot of the confusion about macroeconomics goes away and people can really manage to communicate properly about how it works, which should be what we are interested in understanding. I am writing a book on this subject and would be happy to share it with anyone who is interested in this theoretical approach to our subject.

David Chester (

Comments for this entry have been closed
Return to top