AD, AS, Y, Output Gaps, Cuba, Monopolistic Competition, and Recalculation

They are all related, honest!

Journalists can be excused for speaking as though Aggregate Demand (AD) and output or GDP (Y) were synonyms. Macroeconomists ought to know better, but we are often just as bad as journalists. Macroeconomists never treat Aggregate Supply (AS) as a synonym for Y. The output gap is always the gap between Y and AS, never the gap between Y and AD. The gap between AD and AS is treated as synonymous for the gap between Y and AS.

What's puzzling is not that we make this conceptual mistake; it's that we can get away with it. We can get away with it because we rarely fall into any significant practical error as a result of this conceptual confusion. Why?

The best way to understand first hand that it is a conceptual mistake and that it can lead to error is to go to Cuba. (Or North Korea, I suppose). Better yet, go to Cuba and teach macroeconomics. Watch the looks on the faces of your students when you try to teach any macroeconomic theory that assumes that output is demand-determined. Because in Cuba it isn't; it's supply-determined (except on the black market, etc.). There is always excess demand. The Cuban output gap is the gap between AD and Y, not between AS and Y.

AD is the (aggregate) quantity of output that people (and firms, and government) are willing to buy; AS is the quantity of output that people (etc.) are willing to sell; Y is the quantity of output that is actually bought and sold. Even leaving aside the distinction between a curve and a point on the curve, we would never let first-year micro students get away with failing to distinguish these three concepts. If you impose a binding price ceiling on a competitive market, the quantity actually bought and sold will equal the quantity supplied, not the quantity demanded. Q=min{Qd,Qs} in micro. But macroeconomists speak as if Q=Qd.

Why is it different in macro (leaving aside Cuba, and perhaps WW2)?

The biggest and best innovation in macro in the last quarter century is that perfect competition got replaced by monopolistic competition as the standard assumption for applied macro models. (Yes, it did take half a century for monopolistic competition to make the transition to macro, but never mind that.) One of the reasons this was the biggest and best innovation is that it let us make sense of this puzzle.

Start with a perfectly competitive market. If prices are perfectly flexible, then quantity demanded, quantity supplied, and quantity actually bought and sold, are always equal. Sticky prices break that Holy Trinity, but if price is on average equal to the competitive equilibrium, half the time price should be above equilibrium and output should be demand-determined, and half the time price should be below equilibrium and output should be supply-determined. So how come it seems to be nearly always demand-determined in market economies?

Start instead with monopolistic competition, where each firm faces a downward-sloping demand curve. Start in equilibrium, where the firm maximises profits by setting output where Marginal Cost = Marginal Revenue, and sets a price above MC. Now hold the price temporarily fixed, and shift the MC curve, or the demand curve. Unless the MC curve shifts up by a large enough amount that MC>P at the existing level of output, it will have no effect on output. And unless the demand curve shifts right by a large enough amount that it would mean MC>P, firms will adjust output to whatever quantity is demanded. Sure, the firm would want to adjust price in response to shifts in the demand or MC curves, but if it can't adjust price (for some reason), the firms wants to sell as much output as it can; it will only ration sales if sales increase past the point where MC=P. Until it hits the MC=P limit, a monopolistically competitive firm is demand-constrained. Within that limit, the MC curve only matters because it influences the price the firm will set when it does adjust price, and that price will influence quantity demanded.

That's why marketing in Canada means trying to find a willing buyer for the stuff you want to sell, while in Cuba it means trying to find a willing seller for the stuff you want to buy.

When we aggregate up over an economy composed of monopolistically competitive firms, we get the same result (at least qualitatively). (And it's the same if there's monopoly power in the labour market too.)

In a perfectly competitive economy, the vertical Long Run Aggregate Supply curve plays two roles: it tells us where the economy will adjust to, in the long run, if prices eventually adjust both up and down. And it acts as a supply constraint. If we start in equilibrium, on the LRAS curve, hold prices temporarily fixed, and increase AD, output will not respond. Competitive firms are already producing where MC=P, and don't want to expand output at that price, even if demand increases. We get rationing instead. AS=Y<AD.

In a monopolistically competitive economy, there are really two LRAS curves, a separate curve for each of those two roles. The first LRAS curve tells us where the economy will adjust to, in the long run, if prices eventually adjust up or down. The second LRAS curve, at a higher level of output, acts as a supply constraint. It's the level of output at which firms would find MC=P, and would not wish to expand output further. Unless there's a very big increase in AD, and prices do not adjust quickly enough, we never hit that second LRAS curve (except perhaps WW2?). The economy is nearly always demand-constrained in the short run. Treating AD and Y as synonyms, while conceptually wrong, works fine in practice. On average the economy is on LRAS(1), but is nearly always to the left of LRAS(2). In terms of LRAS(2), market economies are almost never supply-constrained.

If the economy were perfectly competitive, "potential output" really would mean maximum potential output; an increase in AD beyond "potential output" would have no effect on actual output, even with short-run sticky prices. In monopolistic competition, "potential output" can mean LRAS(1). Only if there's a really big shock to AD, so that it gets beyond LRAS(2), do we really hit the wall of maximum potential output.

Recalculation? Right!

The Cuban economy seems always to be facing a major recalculation problem. The disruption of trade when the Soviet bloc collapsed caused a massive recession. Yet there was no shortage of AD. Cubans wanted to buy more goods, but they couldn't find anything to buy. The "monetary overhang", or excess supply of the medium of exchange (as evidenced by a tenfold depreciation of the peso against the dollar) was the flip-side of the excess of AD over Y.

One reason why a recalculation theory of macroeconomic fluctuations makes sense in Cuba is obvious. It's straight Hayek/von Mises. Rather than a price system trying to do the re-calculating, you have a central planner, trying to tell peasants where the best place is to grow coffee. And the changing ideas of the central planner are just one more shock to which the Cuban economy must adjust.

But a second reason is less obvious. The Cuban economy is always supply-constrained, so any shock to AS has an immediate impact on Y. Recalculation is an AS shock.

Suppose the Canadian economy were perfectly competitive, rather than monopolistically competitive. Start in long run equilibrium. Now suppose there's a shock to relative demand. Half the firms get a 10% decrease in demand, and the other half get a 10% increase in demand. No change in Aggregate Demand. Hold prices temporarily fixed. Half the firms cut output 10%, the other half hold output constant and ration sales. On average, output declines by 5%. We see a recession.

Now repeat the experiment assuming monopolistic competition. Half the firms cut output 10%; the other half increase output 10% (assuming MC<P). On average, output stays the same. No recession.

It doesn't end there, of course. In the longer run, firms with declining demand will cut their prices relative to firms with rising demand, and resources will flow from the former to the latter. And those flows will involve some real costs, and might involve higher than normal unemployment rates during the adjustment. There is such a thing as structural unemployment.

But monopolistic competition might explain why AD shocks seem to matter much more than recalculation for market economies.

P.S.1: This post is sort of a response to Scott Sumner, Tyler Cowen, and Arnold Kling.

P.S.2: I wanted to post some diagrams to help readers understand macroeconomics with monopolistically competitive firms. Yesterday, my daughter even taught me how to use "Paint". But TypePad doesn't seem to recognise Paint. Sorry.

Comments

[Adam P]

Nick, I've only just scanned it, it's pretty long, but my first thought is: very good post.

[Nick Rowe]

Thanks Adam! Unlike some of the other stuff, macro with monopolistically competitive firms is actually a topic on which I've got my intuition more or less straight. Really wish I could figure out how to draw and post the diagrams, because I have some neat ones that I haven't seen in the literature. Will maybe do a second post on the diagrams, when I do.

[Stephen Gordon]

If you can get MS Paint to generate image files (,jpeg, .png, .pdf, etc), you should be able to post them.

[Just visiting from macleans]

Or draw it by hand, and scan it into jpeg etc. The hand need not always be invisible.

[Leigh Caldwell]

Very interesting analysis.

The dynamics of a real economy are diverse, of course. Some sectors are closer to perfect competition while others are monopolistic; and where there is monopolistic competition, there may not be as much as 10% (in your example) gap between current and potential output. Even if there is, there will still be a time lag between seeing extra demand and being able to satisfy it.

This is especially true if the extra demand is for savings (that is, for investment). The delay in being able to identify new investment opportunities and lend to them is what creates the excess demand for the medium of exchange which you often mention. Your theory of loan officers from a couple of weeks ago I think expresses this time lag quite well.

Perhaps these lags are equivalent to the recalculation argument - with monopolistic competition and sticky prices being specific causes of the lags, rather than being the fundamental cause of recessions in themselves.

p.s. feel free to email me your Paint files if you want help in converting them to PNGs and uploading - but you should be able to use "Save As" from MS Paint as Stephen suggests.

[Alex Tabarrok]

For really nice graphs get Mayura draw and then save in a variety of formats.

[jsalvati]

This program looked kind of nice to me for drawing teaching graphs, and there's a free trial. I am surprised it is hard to add graphs. Seems like there would be an "insert image" button wherever you write your post; that's what most other blog services have.

http://www.omnigroup.com/applications/omnigraphsketcher/

[Frances Woolley]

On paint, click 'save as' and try one of the formats Steve suggested. Or go into the print menu, and in that box towards the top of the menu, see if it allows you to choose Adobe acrobat .pdf or some such as a printer. I think you can probably do that on your work computer, anyways.

[Rogue]

you might try uploading your graphs at google docs, then post a link to the page that contains a specific graph. Haven't tried it but the advantage over uploading a pdf i think is that you can edit it in realtime even after uploading. you can tweak and edit without having to re-upload each time, and your links will always display your latest tweaks.

[www.facebook.com/profile.php?id=611985302]

So what happens when the goods and services market is monopolistic and the labour market is monopsonistic?

My guess is that labour market issues drive AS down. There is much trying to paper over the cracks but eventually recessions get triggered.

[Winslow R.]

To be 'fair' to Cuba, I've heard different over the last year from my friend Manuelo who would not move back as he is a healthy, young worker in his prime. His story matches this link
http://www.mahalo.com/cuba-wage-caps

He says working Cubans receive a small peso monthly wage, approx 20 pesos with which to buy 'extras'.

Basic services are free. Rent, health care, education etc. AS exceeds AD. Definitely not true in U.S. the 'market economy' given our masses of homeless, sick and uneducated.

Vouchers come each month for a certain amount of nutrition. People that are inclined to eat more go hungry unless they have a way to barter/foreign exchange for more food and this tends to keep people skinny.

Purchases of 'luxuries' requires currency. AD exceed AS.

Your post makes it seem like no one wants to 'net save' in Cuban Pesos. Does everyone in Cuba have money but they have no goods to buy? I don't believe that is true or else how could Cuba peg their currency to the dollar? Do you have black-market exchange rates?

[Too Much Fed]

I'm want to see the graphs!

[cushion]

Great blog. I am an economics graduate - now an economics consultant.

i've written many articles on my blog relating to economics too. Please check it out and follow me:

http://windowsillseat.blogspot.com/search/label/Economics

[www.facebook.com/profile.php?id=611985302]

Cuba runs 2 currencies. Cuban pesos and convertible pesos. Cuban pesos are officially not supposed to be converted. There is an official exchange rate (Cuban<=>convertible), the rate on the street is higher than that.

To make money in Cuba you work in the tourist trade, that includes renting rooms and running local restaurants.

[Adam P]

Leigh, this statement, I think, is wrong: "This is especially true if the extra demand is for savings (that is, for investment). The delay in being able to identify new investment opportunities and lend to them is what creates the excess demand for the medium of exchange which you often mention. "

There is a good reason why variation in investment drives the business cycle but it has nothing to do with the delay's in identifying investment opportunities.

Start with an economy in equilibrium at full-employment, this means that the current real interest rate is equal to the natural interest rate and satisfies two conditions:

1) consumption Euler equations are satisfied.
2) the risk-adjusted expected return on the marginal investment project equals the real rate.

Now suppose a shock drives us out of equilibrium, in particular suppose the natural rate falls or the real rate rises, either way so long as they are no longer equal and the real rate is too high.

In principle consumption can fall immediately, people just stop buying stuff, and conumption falls until it is low enough relative to future expected consumption so that the Euler equation is again satisfied.

Investment falls immediately too, but this does not raise the marginal product of capital because the marginal product of capital depends on the level of the capital stock. The marginal product of captial rises only slowly as depreciation reduces the capital stock. Thus investment falls further than consumption and stays depressed longer.

In the absence of a monetary response to bring the real rate down the recession won't end until the capital stock falls to the point that the marginal product of capital, and hence the natural interest rate, again equals the real rate, at which point full-employment is restored.

[Nick Rowe]

Thanks everyone for the comments, and the helpful advice on how to do graphs. Sorry I've been so slow responding. Yesterday afternoon was very abnormal. I had to do an unscheduled drive to Toronto and back (didn't get home till 3.30 this morning) through snow and high winds (and dealing with total transmission failure on daughter's car). So now I'm catching up on sleep, and preparing for lectures tomorrow morning.

To continue Adam's point above:....which explains why investment demand is so very sensitive (elastic) with respect to the rate of interest, expectations, and everything else.

Will return later. (And with your help I have figured out now how to post paint, but need to practice painting good diagrams.)

[Jon]

Nick: Suppose a bank increases the provision of credit. Under your assumptions AD shifts. It continues to do so until P=MC. At this point the economy becomes supply constrained and the recalculation theory applies.

Consequently, I would conclude that your analysis demonstrates a much weaker conclusion that you submit. Namely, that a modern market economy can tolerate a certain amount of credit expansion; however, I don't think your conclusion holds for a very large credit expansion.

Moreover...
Mises's description of the credit cycle depends on the credit expansion lowering the real rate; however, if P always rises faster than the slope of the MC curve this will never occur. I.e., the expansion must generate inflation only. But if real-rates do shift (owing to a scarcity of information, sticky prices, whatever) this also implies that the P rises more slowly than the MC curve. Which in turn leads to the supply-constrained situation you describe.

I'd say you've inadvertently made a very strong case in favor of Mises's narrative being possible.

[scott sumner]

Excellent post. Unfortunately it's late and I need to think about it for a while.

[Too Much Fed]

Nick said: "Start instead with monopolistic competition, where each firm faces a downward-sloping demand curve. Start in equilibrium, where the firm maximises profits by setting output where Marginal Cost = Marginal Revenue, and sets a price above MC."

From a stock perspective, is that equivalent to gross margin and revenue maximization?

Also, what scenario(s) lead to cost minimization and output, pricing, and gross margin maximization?

[Too Much Fed]

Nick said: "In a monopolistically competitive economy, there are really two LRAS curves, a separate curve for each of those two roles. The first LRAS curve tells us where the economy will adjust to, in the long run, if prices eventually adjust up or down. The second LRAS curve, at a higher level of output, acts as a supply constraint."

Does the first LRAS have a lower level of employment?

[Nick Rowe]

Finally I find some time to respond to comments:

Too much Fed: "Does the first LRAS have a lower level of employment?"
Yes. Lower employment and output (unless the labour supply curve is perfectly inelastic, and wages are perfectly flexible).

"From a stock perspective, is that equivalent to gross margin and revenue maximization?".
No. Firms maximise profits, which is total revenue minus total costs. The "gross margin" (if that means the markup of price over marginal cost) is determined by the elasticity of demand. It's determined by: P = MC/(1-1/E) where E=elasticity of demand, IIRC. So in the limit, as E goes to infinity, as under perfect competition, we get P=MC.

Scott: Thanks! There's no way I can keep up with your rate of producing great posts.

Jon: "Suppose a bank increases the provision of credit. Under your assumptions AD shifts. It continues to do so until P=MC. At this point the economy becomes supply constrained and the recalculation theory applies."

Agreed. Unless prices increase first, which they probably will, unless it's a very big and very fast rightward shift in the AD curve. But we don't normally see cases in market economies where people want to buy more stuff than firms are willing to sell. Rationing of buyers happens, (and you can always find a few special cases), but it seems to be rare, though less rare in a boom. WW2 (in the UK anyway, more so than in Canada?), when there was a very big increase in AD and prices were held down, is the exception that proves the rule. Normal booms don't approach the supply constraint for most goods. (That's my sense, anyway.)

Back to practicing with Paint!