Monday, February 24, 2014

My Quiz for Wannabe Keynesians

Simon Wren-Lewis has a great post today on what makes a Keynesian.  Here is my answer together with a quiz for wannabe Keynesians.

First, let me delve into a little highbrow theory.

Figure 1: The Keynesian Cross

Figure 1 is a picture that goes by the name of the Keynesian cross.  On the horizontal axis is income; the value of all wages, rents and profits earned from producing goods and services in a given year.  On the vertical axis is planned expenditure; the value of all spending on goods and services produced in the economy in a given year. Since this is a closed economy, all expenditure is allocated to one of three categories; expenditure on consumption goods, expenditure on investment goods and government purchases.  Since every dollar spent must generate income for someone; in a Keynesian equilibrium, income must equal planned expenditure.

The upward sloping green line, at 45 degrees to the origin, is the Keynesian aggregate supply curve. This green line is the Keynesian theory of aggregate supply.  It says that whatever is demanded will be supplied.

The upward sloping red line is the Keynesian theory of aggregate demand.  In its simplest form, G and I, represent exogenous spending by government and by investors.  The idea that investment is exogenous, was Keynes' way of closing the system.  He thought that investment is driven by the animal spirits of investors.

The Keynesian model says, that in equilibrium, the economy will come to rest at a point where the green line and the red line cross.  There is  no necessary reason why that point should be associated with full employment, and most of the time, it won't be: Hence, the title of Keynes' book, the General Theory of Employment Interest and Money.

The jewel in the crown of Keynesian theory is the consumption function, represented by the equation,

C = A + bY,

where A is autonomous consumption spending (this is a constant) and b is the marginal propensity to consume (this is a number between zero and one). 
Figure 2: The Great Depression
Figure 2 is the Keynesian explanation for the Great Depression. The top red line, labelled X1, represents expenditure in 1929 before the stock market crash. Government purchases were equal to G and investment was equal to I1.

The lower red line, labelled X2, represents expenditure in 1932, after the stock market crash. Government purchases were still equal to G, but investment fell from I1 to I2.  This shifted down the Keynesian aggregate demand curve and led to a drop in income that was bigger than the original drop in I. Expenditure, equal to income, came to rest at point Y2.  The fall in Y was bigger than the fall in I because, as investment fell, so consumption also fell. And every one dollar of reduced consumption led to an additional drop in income of b dollars.  That at least is the theory.   

Keynes' remedy? Government must spend to replace the lost investment spending.
Figure 3: The Keynesian Remedy
Figure 3 shows how that is supposed to work. The lower dashed red line is aggregate demand in 1932 in the depth of the Depression. Government purchases did not change much during the 1930s, but as the world entered WWII, government purchases in the U.S. increased from 15% of the economy to 50% in the space of three years. As G increased from G1 to G2, the dashed red line on Figure3  shifted up and expenditure went from X2 back up to X1. Notice that  G2 + I2 on Figure 3 equals I1 + G on Figure 2.  

The increase in the size of government in war time was huge and was enough to restore full employment at Y1; but now the spending that had been carried out in 1929 by the private sector was  carried out in 1943 by the government.  We cured the unemployment problem by putting all of those unemployed people in the army.

It is also worth pointing out that private consumption expenditure fell during WWII. Keynesian theory predicts that it should have increased.  That fact is a bit uncomfortable for textbook Keynesians who appeal to the special circumstances of a wartime economy.  Here's Tyler Cowen's take on that debate.

OK: enough theory.  Here is My Quiz for Wannabe Keynesians.

1: Does demand determine employment?  Is the 45 degree line an aggregate supply curve?  And if you answered yes to this question: Is the 45 degree line a theory of aggregate supply in the short run, or in the long run?

2: Does you answer to (1) depend on the assumption that prices are sticky?

3: Is the Keynesian consumption function a good way to think about aggregate demand?  Does consumption depend on income, and if so, what is the value of the marginal propensity to consume? 

If you answered YES to all three questions, you are a bonafide card carrying Keynesian of the New York Times variety.  

Here are: My answers to My Quiz for Wannabe Keynesians

1).  YES.  The 45 degree line is an aggregate supply curve.  Further, it is a LONG RUN aggregate supply curve.  Forget about a vertical Phillips curve.  It simply is not there in the data.

2).  NO. The fact that anything demanded will be supplied has absolutely nothing to do with sticky prices or wages.  I repeat; the 45 degree line is a LONG RUN aggregate supply curve.  Prices and wages are determined by monetary factors and by beliefs and there is little or no evidence that they adjust to clear markets as classical theorists would tell us.

3).  NO.  The Keynesian theory of consumption is not a good theory of aggregate demand. The evidence for a large multiplier is weak or non existent.  I am perfectly willing to be proved wrong on that point; but I have kept up with all of the available evidence (nicely summarized here by James Hanley) and I do not find it convincing. 

Valery Ramey's work suggests that the multiplier, if anything, is slightly less than 1 and she finds that most cases private spending falls significantly in response to an increase in government spending.
Perhaps things are different at the lower bound?  Nope! Ramey and Zubairy
...find no evidence that multipliers are different across states, whether defined by the amount of slack in the economy or whether interest rates are near the zero lower bound.
I am not a right wing market touting Chicago card carrying loyalist.  I want to find evidence to support effective policies to combat recessions.  But I am tired of listening to diatribes arguing that there is incontrovertible evidence that fiscal policy is effective at the lower bound. There isn't.

Am I a Keynesian? I believe I am; but you can judge for yourself.  My work explains (1). I am skeptical about (2) because the evidence suggests no stable connection between unemployment and inflation.  In my view ANY combination of unemployment and inflation can hold in a steady state equilibrium. The assertion that we must cure unemployment with traditional fiscal policy is not supported by the facts and arises from a doctrinaire approach to the evidence.

The challenge for macroeconomic theory is to understand how fluctuations in asset markets are transmitted to aggregate demand and the fact that I am skeptical about the effectiveness of traditional fiscal policy does not mean that I think that the market should be left to correct itself. There are alternative policies we can try.  But that is a story for another day.


  1. I too would answer yes, no, no. But I am a pretty ardent supporter of fiscal stimulus. My arguments for fiscal stimulus (see here — are not multiplier dependent, just that it's a way of using up slack capacity to ease the suffering (and loss of skills, and degradation of industry, etc) of the unemployed and that it supplies a shock that can jog the economy away from a bad equilibrium. As soon as the animal spirits recover (which, in the long run, they will) the stimulus can be ceased. In other words, stimulus should not be thought of so much as a cure ("the illness is inadequate aggregate demand, make up the aggregate demand and the problem goes away!") but as an anti-inflammatory medication and psychological placebo that eases the body until natural recovery mechanisms kick in...

  2. FWIW, my answers:

    1 (a) Yes.
    1 (b) No; the 45-degree line is the goods-market equilibrium condition.
    2 No; all that is required is that prices aren't optimally flexible. (Mere volatility isn't enough.)
    3 Any worthwhile model will have a consumption function. The MPC will depend on the precise specification. But other components of the Keynesian framework (liquidity preference and the marginal efficiency of capital schedule) are just as important.

    Not sure whether I'm a bonafide card carrying Keynesian of the New York Times variety. Don't much care.

    1. Kevin
      1 (b) In the GT AD = AS is both a goods market equilibrium condition and a labor market equilibrium condition.
      2. In my view, wages and prices are secondary. Quantities clear markets.
      3. Yes: but thats not the issue. My world has a near Ricardian consumption function. Yes: the Keynesian cross is a simplification. But a useful one.

  3. John
    I'm ok with that. Its certainly not a bad time to be rebuilding infrastructure. But if the multiplier is indeed 1, or even less, as I fear; then the required size of the stimulus to restore full employment is truly massive. I am not confident that government would effectively manage resources on that scale outside of war time. I'm also not confident there IS a natural recovery mechanism.

    1. America has never had a wartime situation on its own soil.

  4. "The 45 degree a long run aggregate supply curve."

    This begs the question of where the limits are. In the short run we can assume that we are far enough below maximum feasible output that we need only contemplate levels of demand that are physically capable of being supplied. In the long run I think there's a problem if animal spirits (or whatever one conceives as driving demand) get so high that they demand the impossible. Surely our resources have limits. (I'm trying to imagine how that could fail to be the case. You can make population growth and productivity growth endogenous, but I still won't believe they can supply enough to make the 45 degree line extend out to infinity. Eventually natural resources become an issue. And in any case at some point the supply of labor and technology must lose its elasticity.) What happens when the impossible is demanded? My reading of Keynes is that inflation happens in that case. Even if the Phillips curve isn't vertical in the long run, it's surely asymptotic to a vertical line (or else it has a kink).

    1. Andy
      Yes; there is an upper bound at full employment. But this is quite different from the AD AS framework that taught in undergraduate classes where there is a second curve relating prices to unemployment, also called an AS curve.

  5. What does 'natural' mean? Isn't the economy an invention of sorts. An act of collective creation reflecting combinations of interacting decisions from various parties, individual and collective?
    Recoveries aren't natural, they are a reflection of policy choices and individual choices all mixing together. Recoveries aren't defined by what has been, or some natural law any more than buildings are. Sure buildings do have to conform broadly to the limits of available building materials, but almost any building can be built. So too, the economy has some basic building blocks, but these blocks can be put together to build any economy we please - or at least any economy that is politically achievable.

    Isn't there a fourth question for any good Keynesian? Don't Keynesian's have to believe we can take action, some action, today; that imagining resolution in some form of stylized perfection, or some distant future is too easy a task? that the fact that something needs to be done means something can be done? or in short the question is, is the economy political or natural?

    1. Dan, my title was tongue in cheek; but yes, Keynes certainly was an advocate for action to design new institutions. So am I.

    2. well, I had in mind a little more than designing institutions.
      This whole question of fiscal stimulus is tricky.

      If C is determined by income; I is determined by animal spirits (or what ever etc.), what determines G?

      Am I missing something, I just find G to be very incompletely described and justified. I wish someone would begin addressing this topic more directly and broadly.

    3. Dan
      G is determined by the political process... typically not modeled by economists. That lets us the question; how can we design institutions that choose G optimally.

  6. "...The upward sloping green line, at 45 degrees to the origin, is the Keynesian aggregate supply curve. This green line is the Keynesian theory of aggregate supply. It says that whatever is demanded will be supplied.

    The upward sloping red line is the Keynesian theory of aggregate demand..."

    This isn't Keynes' AD-AS Model. This is Samuelson's Income-Expenditures Model, often called the Keynesian Cross.

    It is derived from a pair of 1939 papers by Paul Samuelson on the interaction of the multiplier and acceleration principle and first appeared as the analytical core of his 1948 textbook. Whether or not the analysis underlying the Keynesian Cross is to be actually found in Keynes is still a matter that is hotly debated.

    The Keynesian Cross is a model of aggregate demand (AD) under the specific assumptions of a fixed price level and fixed interest rates. The 45 degree line represents the equilibrium condition where income equals expenditures. The other line represents planned expenditures, and is a function of income. Some textbooks label the horizontal axis as output, production or GDP in addition to income, but this is under the assumption that all points on the 45 degree line are also at the point of intersection between AD and AS, which Keynes termed "effective demand" in Chapter 3 of "The General Theory of Employment, Interest and Money". However, in Samuelson's original 1948 textbook, the horizontal axis was simply labeled "national income".

    Key features of Keynes' analysis are completely missing from the Keynesian Cross. It is a model of AD in which prices, interest rates and money play absolutely no role.

    1. Mark
      You are right of course: but the income expenditure model does not do a bad job of capturing the essence of the first few chapters of the GT. Here is a link to my version of the income expenditure model which has the advantage of being internally consistent and not assuming sticky prices.

  7. Roger: my answers:

    1. Demand determines employment in this model. The 45 degree line is definitely not a supply curve. Full equilibrium requires: Ys=Y=Yd. What we have here is semi-equilibrium, where Yd is a function of Y, so only the point at the intersection is a level of Y such that Y=Yd(Y). It says nothing about Ys. A supply curve tells you how much output people *want* to sell.

    2. Not necessarily.

    3. It may be part of a theory, but it is a not a theory. MPC depends on duration and borrowing constraints.

    1. Hi Nick
      1. "The 45 degree line is definitely not a supply curve". Odd you should say that as I recall getting that idea from David Laidler in our first year grad macro class at UWO.

      For my take on AD AS take a look at the piece I linked in my reply to Mark Sadowski. It appears in a theory journal and is quite math intensive: but you can skip that and read the intro and conclusion to get the intuition.

      The bottom line: 1. the 45 degree line is just the first order condition for labor. 2. Fixing the wage is just a choice of numeraire.

    2. Roger: Yep. I too remember David saying the 45 degree line was not a supply curve. And that the downward sloping AD curve in {P,Y} space is not a demand curve either. I think he was right. I interpret the 45 degree line as saying Q=min{Qd,Qs}.

      I had a look at your paper. Heavy duty stuff. But I think the key is that your household only lives for one period, and cannot sell its stock of K (land) by assumption, so cannot dissave, and gets no benefits from saving.

      Let me simplify your model drastically. Assume each household produces haircuts, but cannot cut its own hair, so consumes haircuts produced by other households. The production function is Y=L. There is no land or capital. Labour provides no disutility, but each household can only produce one haircut per day, so Ys=1. (That is the supply curve.) Since your household can neither save nor dissave, it will consume its disposable income. So C=T+(1-t)Y, where T is lump sum transfer and t is tax rate. since there is no I or G, we know that output demanded Yd=C. Assume the short side of the market determines quantity of haircuts traded: Y=min(Yd,Ys}. And there's the model. Adding in search doesn't affect anything. It doesn't matter what they do when they are unemployed and so not cutting hair.

      And if the government sets T=0 and t=0 then any level of Y between 0 and 1 is an equilibrium, because the consumption function coincides with the 45 degree line.

    3. Put it this way:

      If we deleted search, and land, from your model, nothing would change.

      If we allowed households to live for longer than one period, and so save or dissave money or land, and if we allowed households to barter haircuts ("I will buy a haircut from you if and only if you buy a haircut from me") everything would change.

    4. Or: suppose the government sets T>0, and t=0. After all, it has a printing press, so can do helicopter money if it wants. There is no requirement that the government that owns a printing press must balance its budget. What is the equilibrium? The only equilibrium in your model is that p=infinity. Because your households never want to hold any money at the end of the period.

    5. Nick
      We have different recollections of David's lectures. I recall him saying that the 45 degree line IS an aggregate supply curve; we should ask him. In any case; whatever David thinks, in my view the 45 degree line in the income expenditure model IS an aggregate supply curve. Try reading Chapter 3 of thee GT and comparing it with what I say in the piece I linked above.

      "If we deleted search ... nothing would change"

      NOT SO!! That is the point of the AD-AS piece. Think of the Beveridge curve as an isoquant of a search technology. We can fill 100 jobs if 1000 search for 200 vacancies or if 200 workers search for 1000 vacancies. Both are equilibria in my model because there are no price signals to guide the search process.

      "If we allowed households to live longer than one period...."
      That is what I do in my EJ articles here and here and in my book here .

      "... The only equilibrium is p=infinity".
      The model described in those papers is a purely real model. No money. The wage is a numeraire. For the monetary version see my piece in the new Phelps Volume here.

    6. p.s.
      Thanks do much for spending the time to comment. I always enjoy reading your blog.

    7. Roger:

      I think I agree with your interpretation of GT chapter 3. If you start with the labour demand curve for a competitive economy: W/P = MPL, then set W=1 for numeraire, then rearrange stuff, multiply and divide, etc. you get Chapter 3. It is basically the same behavioural equation that generates the textbook SRAS curve in a model that holds W fixed.

      I will check my old lectures notes tomorrow, to see what I thought David said!

      I did a post on my interpretation of the 45 degree line:

    8. The government sets both T (lump-sum transfer) and t (tax rate).

      And we impose the condition that the budget must be balanced in equilibrium. Which means T=tY.

      But then the equilibrium level of Y drops out immediately: Y = T/t.

      Does it make any difference to the model if the unemployed spend their time searching for jobs, digging their gardens, walking the hills, or waiting miserably? As long as they would prefer a job.

      That's why I said I didn't think that search was playing any role in this model.

      Here is a 3 year old post where I tried to build a very simple "Roger Farmer type" model. Not sure if I interpreted you correctly. But it is very simple, and does have search, and money, and microfoundations.

    9. Nick

      "Does it make any difference to the model if the unemployed spend their time searching for jobs, digging their gardens, walking the hills, or waiting miserably? As long as they would prefer a job."

      YES. It matters a lot. Search is an input to the matching function. If you want to complicate the model by adding leisure -- thats fine. But now the participation rate becomes a separate variable.

  8. i would like to see an investment good defined

    1. From the System of National Accounts (SNA, 2008, p. 8): "The activity of gross fixed capital formation, like intermediate consumption, is restricted to institutional units in their capacity as producers, being defined as the value of their acquisitions less disposals of fixed assets. Fixed assets are produced assets (such as machinery, equipment, buildings or other structures) that are used repeatedly or continuously in production over several accounting periods (more than one year). The distinction between intermediate consumption and gross capital formation depends on whether the goods and services involved are completely used up in the accounting period or not. If they are, the use of them is a current transaction recorded as intermediate consumption; if not it is an accumulation transaction recorded in the capital account."

  9. Hi Roger:
    I know you're trying to make a simple point. And have no problem with that. But Keynes actually did believe that flexible prices would make things worse, and did not presume that prices were rigid to get to unemployment equilibrium. That's actually Modigliani in his 1944 paper. Keynes suggested (ch. 19) that income distribution with falling wages and prices might shift against workers with higher propensity to consume and debt deflation would lead to a decline in spending which would overcome any positive effects that deflation would have in investment via lower rates of interest (Keynes' effect). In other words, question 1 and 3 are relevant, but 2 actually makes you a neoclassical Synthesis (or maybe New Keynesian) economist.

    1. Matias

      "Keynes actually did believe that flexible prices would make things worse"

      Yes he did. And yes, that IS Modigliani speaking. What a lot of damage he did.

      The post-Keynesians had a point when Joan Robinson referred to the watered down version of Keynesian economics that came from Cambridge Mass as "bastard Keynesianism". Regrettably; no-one from Cambridge England articulated the post-Keynesian view in a language that neoclassicals could accept.

  10. Hi Roger,

    Very interesting post! I’ve tried to answer your questions in the way Keynes, himself, might have done (without the beautiful prose sadly).

    1. “Does demand determine employment?” It would be more in keeping with the General Theory to say that *expected* demand determines employment. Specifically, aggregate employment is determined by firms, which choose the volume of output, and the quantity of labor, they believe will maximize their profits given their cost estimates and demand forecasts.

    2. “Is the 45 degree line an aggregate supply curve?” The aggregate supply curve (Z in the GT) shows the volume of output that profit-maximizing firms would be willing to produce at varying wage levels, which I don't believe is the same your 45-degree line.

    3. “Is the 45 degree line a theory of aggregate supply in the short run, or in the long run?” If, by “long run,” you mean a period long enough to adjust the capital stock, then the answer is “no,” because Keynes was doing a short period analysis.

    4. “Does your answer to (1) depend on the assumption that prices are sticky?” Not for Keynes, who assumed that prices and wages would rise at higher levels of output (because Keynes also assumed diminishing returns).

    5. “Does consumption depend on income, and if so, what is the value of the marginal propensity to consume?” Yes, C depends on Y, if only because a large number of households live paycheck-to-paycheck. One study estimates an average MPC of 48% (, though I’m not in a position to evaluate this result.

    1. 1. Fair point.

      2. No its not- but there is a transformation that makes it so. Take a look at Figure 1 on page 91.

      3. I mean "are there forces in play that will retire a classical equilibrium?" My answer: NO.

      4. A+

      5. In my view, permanent income, or wealth, is a more important determinate of consumption.

    2. I agree with you on #3, and so would Keynes. Regarding #5, survey data show that about 40% of Americans are living paycheck-to-paycheck, with 47 million relying on food stamps. Why not add a bit of heterogeneity here, distinguishing between households that spend all, or almost all, of any small addition to their income, and households whose spending is explained by permanent income?

  11. These free economics class notes may be of interest.
    I just added a link to this article in

  12. "He thought that investment is driven by the animal spirits of investors." Did Keynes really think that? My impression is that he thought that investors increase supply in the face of high demand. The orthodox theory of economics seems to assume animal spirits cause investors to produce goods that no one has the money to buy. Keynes relied on simple business sense rather than appeals to mysticism.

    I think it simply shows ignorance to argue that falling C during WWII shows that G always crowds out C. Remember that during WWII, the government seized complete control of the economy. You couldn't buy sugar, gasoline, tires, nails, boards or just about anything else without clearing it with some government board or another. The economy was running flat out under detailed government control, and C was simply cut by government fiat in the face of rising G for the war effort. A new typewriter ribbon for your office might mean one less typewriter ribbon writing out orders on the front. When I have to go to a government website to authorize my next purchase at Amazon, I'll believe G crowds out C.

    My answers:

    1a) Demand does determine employment subject to the ability to raise productivity and quantization effects. That is, during periods of economic growth, it can be hard to hire fractional employees.

    1b) Yes. The 45 degree line is an accounting identity in the long run. Money has to go somewhere, even if it just goes to buy symbolic items like treasury notes or Keith Haring prints. Of course, if more money is being printed or money is being withdrawn from circulation, then the slope may be different in the short run.

    2) My answer has nothing to do with prices or wages, sticky or otherwise. I just assume that the pigeon hole hypothesis applies to money just the way it applies to everything else.

    3) The Keynesian theory is an excellent theory of aggregate demand. It argues that money for buying things has to come from somewhere and, once spent, it has to go to somewhere. It's sort of a conservation law for money. Too many economic theories make no accounting sense because they violate this rule.

    Also, don't be too quick to dismiss the possibility of a large multiplier. It's one of the reasons high taxation is associated with high economic growth and, more particularly, with rising standards of living. Every time the government steps up spending, it seems we have some kind of boom. Look at the Civil War boom which established so many industries. Look at WWI which led to the Roaring 20s. Look at WWII which led to one of the longest periods of economic growth in our nation's history. Why do you think the Republicans are so big on borrowing and spending? Why are they so serious about "austerity" when a Democrat is in office? They may deny it loudly, but they are all Keynsians now.

    1. Kaleberg
      "Did Keynes really think that ... [investment is driven by the animal spirits of investors.]?" Judge for yourself:

      "Most, probably, of our decisions to do something positive, the full consequences of which will be drawn out over many days to come, can only be taken as a result of animal spirits—of a spontaneous urge to action rather than inaction, and not as the outcome of a weighted average of quantitative benefits multiplied by quantitative probabilities." GT Ch VII.

      I think it simply shows ignorance to argue that...

      "... falling C during WWII shows that G always crowds out C." Did I say that somewhere? Nope...

      "Also, don't be too quick to dismiss the possibility of a large multiplier. It's one of the reasons high taxation is associated with high economic growth "

      Interesting point of view. Lets raise taxes.

  13. I argue that the "Phillips curve" is predominantly a property of the unemployment rate: it tends to rise quickly and fall slowly with an almost predictable slope. Therefore most of the time series data comes from periods of falling unemployment (recovery, economic growth and a general rise in the price level).

    The Phillips Curve

    (There appears to be a stable relationship between inflation and a more complicated function of unemployment and its derivative that masquerades as the Phillips curve because for most of the time series data the derivative < 0.)

  14. The macroeconomic response in the US to World War II is actually much subtler than you indicate here, and (in my view) much more amenable to Keynesian conclusions.

    Following the Nazi invasion of Western Europe in May 1940, the US began a large-scale military buildup, even though it was not yet officially in the war. This is visible in the national accounts (annual, since quarterly data is not available for this era): there was a 74% increase in real national defense spending from 1939 to 1940, and a spectacular 420% increase in real national defense spending from 1940 to 1941. The overall effect was to push up national defense as a share of GDP from 1.8% in 1939 to 11.9% in 1941, already a very large impulse. (Though, of course, not nearly as large as the increase to come, once the US actually entered the war.)

    What was the effect on nondefense spending? It did quite well, with no visible crowdout whatsoever. Real personal consumption expenditures had increased by 5.6% from 1938 to 1939, and they increased by 5.2% and 7.1% from 1939 to 1940 and 1940 to 1941, respectively. The 1940 to 1941 interval is the most remarkable: despite the 420% increase in national defense spending, durable goods PCE increased by a robust 15.4%! Indeed, despite the vastly increased allocation to military expeditures, literally every component of private demand increased in both 1940 and 1941 - durable PCE, nondurable PCE, services PCE, nonresidential structures investment, nonresidential equipment, residential investment, you name it. This certainly suggests a multiplier greater than 1.

    Private consumption and investment only decreased starting in 1942, once the US entered the war and built up military spending even further. This is no surprise: at some point, capacity constraints will be reached, and any further expansion without monetary offset will result in runaway inflation. (Of course, what actually happened during was that inflation during the war was artificially bottled up by price controls and rationing.) Even the most extreme Keynesian believes this - the question is, starting from a given amount of slack, *when* it will happen. And the real GDP data (which I am drawing from NIPA table 1.1.1) seems to indicate quite clearly that this point was only reached *after* a buildup that had already increased defense spending by 10 percentage points of GDP, far larger than any stimulus program ever contemplated in the postwar era.

  15. By the way, in the US there are only two large-scale war episodes for which we have GDP data: World War II, which increased military spending as a fraction of GDP by about 40 percentage points; and the Korean war, which led to a increase of about 9 percentage points. All other wars, including Vietnam, led to an increase of 2 percentage points or less, and from a macroeconomic perspective were (relatively) negligible.

    What happened in these two wars? As I already mentioned, World War II followed a playbook quite consistent with the Keynesian framework: the initial surge in military spending increased private spending as well, but eventually military spending grew high enough to caused crowdout.

    The Korean war is also friendly to the Keynesian story, although not overwhelmingly so. Despite the large increase in military spending, *every* component of private demand except residential investment increased from the prewar 1950Q2 to the peak defense spending 1953Q2 immediately before the armistice. Real PCE increased at a 3.8% annual rate, which is identical to the 3.8% trend growth rate in real PCE over the 20 years following the start of quarterly national accounts in 1947. The performance of private investment over the period was a little worse, with 1.8% average annual growth - this is due to the constraction in residential investment, which happened to be in a very large boom immediately before the start of the war and was hurt by the Fed-engineered interest rate hike.

    Furthermore, following the war there was a slowdown or outright contraction in most components of private demand, leading to the first NBER recession of the 50s. From 1953Q2 to 1954Q2, PCE inched up by only 0.8%, well below the rate at which it increased during the war. Durable goods PCE actually fell outright by 1.9%, and private investment fell by a spectacular 10.0%, due to large falls in equipment and inventory investment (which overwhelmed the increases in structures). This came despite a 12.5% decrease in defense spending.

    Thus the Korean war, while not as dramatically supportive of a Keynesian model, hardly indicates a multiplier well below 1: continued private demand growth from prewar to the war's peak indicates a multiplier of about 1, while the fall in private demand at the end of the war indicates a multiplier of above 1. My guess is that these events reflect the fact that

    (1) there was relatively little slack at the beginning of the war (the economy was in a healthy rebound from the 1948-1949 recession),

    (2) the war was primarily financed by an increase in taxes (the budget was roughly balanced throughout the Korean war), dampening or eliminating the traditional Keynesian multiplier, and

    (3) the Fed continued to raise interest rates throughout the war to restrain private demand, and then failed to cut interest rates quickly enough when the war ended to prevent a brief but nontrivial recession.

    1. Thanks for your very detailed account of the history of expenditure. We still have the issue of identification to deal with. Here is a paper by Ramey and Zubairy that gives me pause for thought. To quote from the abstract:

      "This paper investigates whether U.S. government spending multipliers differ according to two potentially important features of the economy: (1) the amount of slack and (2) whether interest rates are near the zero lower bound. We shed light on these questions by analyzing new quarterly historical U.S. data covering multiple large wars and deep recessions. We estimate a state-dependent model in which impulse responses and multipliers depend on the average dynamics of the economy in each state. We find no evidence that multipliers are different across states, whether defined by the amount of slack in the economy or whether interest rates are near the zero lower bound. We show that our results are robust to many alternative specifications. Our results imply that, contrary to recent conjecture, government spending multipliers were not necessarily higher than average during the Great Recession."

      I am aware that other researchers make different claims.

  16. This comment has been removed by the author.

  17. Figure 1 works as a nomogram (or nomograph). It demonstrates that expenditure calculated with one formula equals income calculated with a second formula. The two formulas yield equal results at the 45 degree line for all values.

    I have problems with the overlay equation(s) X = A + I + G +bY because of the way it behaves when Y approaches zero or becomes negative. It WOULD make sense to me if Y were considered accumulated money supply (capital?) but then Y would not be income.

    I can see that Keynes and others are suggesting that more income would result in more spending. I think that is correct.. Unfortunately, the display in Figure 1 does not logically translate well.

    I would suggest fixing the logic by labeling the Y axis as "Y plus Annual Money Supply Change (AMSC)". The 45 degree line would than represent two equation equality, as it does now. The equation
    "X = A + I + G + b*(Y + AMSC)" would make a lot of sense.

  18. Roger, forgetting all the discussion about what did Keynes mean or not and getting back to your version of keynesian economics.
    1) If the government reduces labour taxes on firms subsidised by a lump sum tax, does it raise employment in your model? In the 1st generation of RBC models with indeterminacy employment would fall (same if productivity increases)- see Aiygari's critique
    2) Can you distinguish your model with standard data sets from a version of Angeletos' and Lao's model of sentiment driien business cycles which in the baseline version at least has a pareto optimal equilibrium? Same question for the sentiment model of your Benhabib et al

    1. daniels:

      These are great questions.

      1) You can take the supply side of my model and add your preferred theory of aggregate demand. My contribution in recent books and papers was to explain the Keynesian theory of aggregate supply. Main conclusion; the 45 degree line in the income expenditure model is an AS curve.

      That begs the question; what determines aggregate demand. Initially, as you will see from reading my piece on AD and AS in IJET, I believed that the model would justify a large fiscal expansion. My views evolved as my reading of the evidence suggests that the multiplier is at best around 1. That comes down to the question: If G increases; how much will AD increase?

      This is very different from the 1st generation indeterminacy models which relied on a classical theory of demand and supply in the labor market. See my survey on the difference between 1st and 2nd generation models of indeterminacy here.

      2) Another great question. Probably not. For any model where the equilibria are Pareto suboptimal, there will always be an alternative where the equilibrium is efficient.

      For example, I am working now on why the stochastic discount factor is volatile. My answer is that sunspot fluctuations cause a Pareto inefficient reallocation of resources between agents. There is an alternative explanation; the discount rate of the representative agent is a persistent random variable that moves around over time. I think my explanation is more plausible; but there will always be holdouts who are wedded to the idea that equilibrium is tautologically optimal.

    2. You can take the model of Angeletos and Lao and get large inefficiencies if for example sentiment shocks affect interbank markets and the resaleability of collateral, so a negative sentiment shock could explain tighter financing constraints (now often obtained in models via direct exogenous shocks to loan to value ratios or monitoring costs), or you could even have sentiment shocks about the level of idiosyncratic uncertainty shocks facing firms.
      Also Bob Hall has recently emphasized how in models with matching frictions and dynamic labour demand, the discount factor volatility directly impacts demand for labour, since hiring is like an investment decision.
      I'm still tending to prefer a theory where sentiment shocks a la Angeletos/Lao + maybe some systematic deviations from rational expectations towards greater pessimisn in recessions (which look like discount factor movements in rep agent models)+feed back into lower consumer search in product markets and lower marketing efforts by firms in recessions which reduces the matching efficiency of product markets explain big recessions without relying on the indeterminacy of bargaining your models are relying on (why would firms take prices and wages as given if there are significant gains from trade? Why not directed search+price/wage posting as a way to eliminate the indeterminacy).
      But I'll keep an eye on your models and consider giving them a nontrivial weight in my conditional forecasts.

  19. Thanks for the reply Roger. One question, and I know classifications and schools of thought are often imperfect, but your views, in which there is no tendency for the
    economy to converge to a long-run natural rate of unemployment, and the latter can be anything depending on the confidence of investors resembles a lot Shackle's views of Keynes. Is that fair to say? And if so, how does it differ from certain Post Keynesian views?

    1. I admit to not having read Shackle. Do you have a recommended source?

      As for the Post-Keynesians; the Post-Keynesians that I have talked to are in broad agreement with me. The main difference is that I accept neoclassical tools.

    2. Matias will have his own list, but I recommend G.L.S. Shackle, "The Years of High Theory," which is a history of economic theory between the wars, and "Keynesian Kaleidics," which is a very short book about the essence of Keynes's economics from Shackle's point of view. For what it's worth, here's my take:

    3. There are many books by Shackle, since we was a prolific writer. My preferred one is really a history of thought one, The Years of High Theory, but the one that would make more sense in this context would be Expectations, Investment, and Income.

    4. Greg and Matias
      Thanks for your suggestions. I have ordered them both.

  20. Thank you for sharing valuable information. Nice post. I enjoyed reading this post. The whole blog is very nice found some good stuff and good information here Thanks..Also visit my page IInvestment Property We are a Registered Tax Agent and have a close working relationship with the Inland Revenue Department (IRD).


Note: Only a member of this blog may post a comment.