« Individual and aggregate, marginal and total, incentives to cut prices in a recession | Main | Proxies for monetary disequilibrium »


Feed You can follow this conversation by subscribing to the comment feed for this post.

The risk of firms exposing themselvess to massive losses when they face bankruptcy is overblown. The incentives rarely lineup well in this way, and only occurs when employees look to much like equity holders, which is rare since most employees act like debt holders. Consider a few points

1. High risk moves tend to blow up quickly especially when compared to the slow slog toward bankruptcy large firms usually face. 4-5 years of cutting costs and rearranging deck chairs means 4-5 years more salary/benefits/pension contribution. It also means 4-5 years of keeping your eye out for a better opportunity while you have your current job.

2. High risk plays that blow up are very public. The more public the disgrace the colder the market will be. Your the guy that tried to turn Borders into a top of the line pedicure franchise? Hahaha, we can't hire you. You are the guy that steered K-Mart into a 10 year demise? Come on board as CFO.

3. The skill set of any decent sized company has been geared toward product X in market Y. Change is difficult and many will resist, and this includes upper management.

US Auto manufacturers are great examples of this. They spent decades slowly losing market share and watching their manufacturing costs kill their competitiveness. By the auto bailout everyone and their mother had an opinion on how to deal with the unions, but management appeared to consistently cave to union demands rather than go for the gusto and try to bust them once and for all. Even on the other end- where big companies reinvent themselves successfully- you don't see high risk strategies. Typically you will find an Apple that went back to its core competency (hey, don't we engineer great products? Lets do more of that) and not a radical likely to fail, but maybe succeed, proposition.

This strategy only makes sense in single iteration games like poker tournements (this is very well understood and even low level book will cover when to go all in based on stack size alone). The few times it is applicable to real business tend to be in financial sectors where bankruptcy is a decision of regulators and the compay won't be allowed to manage its own demise.

So, the problem is with trying to do representative agent macro.

They worker/entrepreneur/creditor/debtors work too much and set prices too high?

Could it be that they would like to work more (so that they won't go bankrupt) and they set prices too high to sell the output because they assume sales will be great.

It is like the barber spends every saturday at the shop (along with monday through friday) and several nights too. But the prices are so high that few customers will buy.

Big supply of labor, high prices, low output.

> Which means that higher real debt would increase the equilibrium level of output.

This is because you've aggregated firms and workers. Firms don't experience leisure, so they cannot choose to work more. Workers do have leisure, so they can work more. Separate these agents, and you return to the intuitive result.

Indebted firms cannot simply choose to work more, since they must buy labour. They still have to maximize the expectation of profit given solvency (such that negative profit has utility zero rather than negative), which leads to higher-than-rational prices and lower output. Seeing an increasing marginal cost for labour would exacerbate the output-reduction effect, as a firm would experience lower input prices for reduced output.


"So there are two effects of increased debt going in opposite directions: Panglossian expectations causing markups to increase and output to fall; and the desire to earn extra income to reduce the risk of going bust causing output to increase."

Loosen your restrictions. Allow a firm to do some combination of 1. Pay off it's debt from the sale of goods and 2. Pay off its debt from the debt repayments it receives.

Because your firms are lenders, the two forces work in conjunction. Firms will increase the markup they charge on the money that they lend to other firms in a desire to earn extra income to reduce the risk of going bust themselves. And so just as the prices of goods will rise under Panglossian expectations, so will the nominal cost of debt repayment.

bacon: I think you have a point. What's the opposite of "Panglossian expectations"? Yep, a manager (unless he has a really big profit-share) will tend to think about the safety of his salary, and act more like a bondholder.

But I was trying to see if I could build a model where everything in Steve's post actually added up right at the general equilibrium level. So I was taking his view about Panglossian expectations as given.

Bill: if we ignore the Panglossian aspect, each individual hairdresser will want to set a lower relative price to increase sales if he wants to earn more income to reduce the risk of default. In macroeconomic equilibrium each chooses a relative price of one, but all work longer hours and sell more haircuts. Given a downward-sloping AD curve, that reduces the nominal price level.

I'm not sure this result would change much if we dropped the representative agent assumption.

Majro: If we disaggregate firms and workers, real wages will fall when individual workers increase their labour supplies to earn more income to reduce risk of going bust, and firms will respond by hiring more labour and selling more output. It makes no difference. Unless you assume sticky wages, which is begging the question.

Majro: but I get your point. It makes a difference if it is firms or households that have the debt.

Frank: the debts all cancel out.

Nick — Thanks for this! I'd have to work very hard and do a lot of research/note-revisiting to generate what you can just throw out as a bog-standard New Keynesian setup.

Note re the question of "why not produce more at lower prices?", I do rig the model in my telling by simply presuming capacity constraints over the period of time over which obligations are due. Whether you model that as a hard limit or just a sharply increasing marginal cost curve (thinking in numeraire, not utils, because numeraire is what must be paid to crreditors), it serves the same function. Note that I don't predict that prices move not at all, only that there will be a floor under a firm's price, set as (firm profits excluding fixed costs at Qmax) / (fixed costs that must be covered to retain control over the firm). So firms usually will drop price somewhat in response to a nominal demand contraction, because they will try to produce and sell more (work harder). But shareholders will refuse to drop the price lower than what would be necessary to cover their nut when producing at full capacity, a constraint which is difficult to loosen over the short horizon over which their obligations come do.

Over time, if firms can renegotiate their cost structure and/or acquire finance to produce at larger scales, their pricing will drift down. This again is a force that drives consolidation and kurtosis of outcomes, firms that have difficult capital structures and/or lose the lottery for unusual marketshare become cheap just when the firms hanging on would really like to expand capacity and reduce price.

(Note also that a worker's ability to expand capacity is much more limited than that of a firm's, a fact that tilts in the direction of labor price stickiness when worker's life arrangements carry difficult to adjust fixed costs.)

Steve: Aha! I'm glad you found it.

"I do rig the model in my telling by simply presuming capacity constraints over the period of time over which obligations are due."

I could easily rig my model like that too. Just drop the disutility of labour, and add a constraint L <= 1. But then the only macroeconomic equilibrium is Y=L=1, regardless of markup of price over marginal cost, because marginal cost is zero until we hit capacity. So we don't get any effect on the price level from Panglossian expectations. The AS curve is vertical at capacity.

" Note that I don't predict that prices move not at all, only that there will be a floor under a firm's price, set as (firm profits excluding fixed costs at Qmax) / (fixed costs that must be covered to retain control over the firm)."

Yep, I think (though I haven't formally proved it) that my model would have that feature, at the macro level. There would be a floor under the nominal price level, and the AS curve would asymptote to that floor (for a smooth probability distribution of the relative demand shock).

I don't understand why these firms with asymmetric incentives develop "Panglossian expectations" about the world as a whole. The are merely motivated to take more risk. Since we are out of perfect competition, taking more risk might mean lowering prices with the Panglossian expectation that you successfully steal away so much market share that gross profits increase. This seems like the problem with such a micro-focused intuition that doesn't seem to tell an equilibrium story. And if I had to fabricate a stylized finance fact from years of watching firms on the brink, I would say their Panglossian behavior tends far more toward this version of desperate discounting than holding the line. Usually it is Bed, Bath and Beyond waiting patiently for Linens and Things to finish its accelerated suicide via price promotion, although there are certainly examples of both kinds of optimism. Some of the most flexible prices show up in the most economically leveraged, clustered, and at-risk industries such as airlines and autos parts, while many stickier prices like candy come from industries who don't have a whiff of existential risk to the marginal, price-setting firm even in recessions.

dlr: assume the individual firm has an upward-sloping MC curve. (That is crucial to my model). For a given elasticity of demand, and so a given profit-maximising markup of price over *any point* on the MC curve, the bigger is demand, the bigger is MC, and so the higher is the profit-maximising price. So if the individual firm thinks its relative demand curve will shift right, it will raise its price. And if the demand curve shifts left, it's bust anyway, so it doesn't matter what price it sets.

On the other hand, if it is *elasticity* of demand that is uncertain (as opposed to level of demand), I don't know what the Panglossian firm would do. Probably toss a coin and either raise or lower price, hoping demand is either less or more elastic than it rationally expected it to be.


"Frank: the debts all cancel out."

But do the firms adopt Panglossian expectations when entering into a lending arrangement - do they charge an interest rate for the money that is lent?

If we want to say that firms adopt Panglossian expectations when setting the prices for goods, then I think we have to expect the same thing of the firms when they lend money out.

Frank: stop. You are wandering off laying out red herrings, again.

dlr: "I don't understand why these firms with asymmetric incentives develop "Panglossian expectations" about the world as a whole."

Well, just to be clear, my firms have perfect information about all macro variables. The only thing they don't know is whether their individual demand curves will shift left or right, in an elasticity-preserving way.

Thinking it over, I think Majro's solution might work. I need to build some sort of chinese wall between the firm and the agent who works there. Then let each agent own the market portfolio of firms' debt and equity. But I don't want to introduce a population of managers, who run the firms, because it would mess up the model.

Frank: This. Model. Is. Not. About. Firms'. Lending. Behaviour. The loans have already taken place, in the past.

And when I ask you to stop commenting with off-topic points you stop. It is not an invitation to argue the point.

Go away.

> I need to build some sort of chinese wall between the firm and the agent who works there. Then let each agent own the market portfolio of firms' debt and equity. But I don't want to introduce a population of managers, who run the firms, because it would mess up the model.

No need for managers, just aggregate profits. Agents earn a per-capita share of net profits from firms, but their capital income is not tied to the profit of any single firm (namely their employing firm). Essentially, everyone invests in an index fund.

Each firm is profit-maximizing, such that they behave in a Panglossian way if short-term solvency is at issue; this behaviour does not need a (potentially rent-seeking) manager-agent.

Majro: but for proper microfoundations, we need to *derive* Panglossian behaviour for the firms. There has to be someone, for each firm, who chooses to set price in that way. It's not strictly kosher to wave our hands and say it would be in manager's interest to set price that way, if there are no managers in the model.

"Shouldn't be hard for a mathematically unchallenged economist (unlike me) to rig something like that up."

Nick, why don't you find a graduate student in math and combine your efforts. Just a thought.

About the desire to produce more in a recession to avoid bankruptcy, my thinking about this problem is a bit different, so I apologize ahead of time if I am off topic:

Suppose your economy has two monopolistically sold goods. Coffee and Tea. Now, what happens when people decide that they no longer like tea but prefer coffee. You expect the tea making firms to shrink (produce less tea) and the coffee making firms to grow (produce more coffee). I would say if the economy does not behave this way, if instead the tea makers try to produce even more tea and sell it at a lower price, then the economy is dysfunctional. Such an economy, if it existed, would not be able to efficiently produce goods that people wanted.

Does it matter how the firms are financed? I can't see why it would. The financing decision merely determines who pulls the plug, not whether the plug is pulled. Let's assume only equity financing, to keep things simple. Equity still has an expected return, so when news of a decreased preference for tea hits, the market re-prices the tea firms downward so that some of the capital is sold off and workers are fired. Similarly, the coffee making firms find themselves more valuable, and are able to purchase more capital and hire more labor.

Now, if this is how the managers of firms react -- by downsizing their capital stock and hiring when demand falls, and increasing their capital stock and hiring when demand increases, then what is the justification for throwing this reaction function out the window when there is a general decrease in demand for both tea and coffee? Wont the managers of both firms continue to follow their decision making process and have both firms downsize?

If not, why would the managers of each firm choose to produce more in a general downturn but not produce more in a downturn specific to their industry, given that managers only care about the price signals sent to their own industry?

I'm not clear on why you're giving so much attention to sticky-or-not goods and services prices. Are you being completist by chasing down even the small issues, or do you think it a big issue relative to the elephant in the room, sticky wages?

Alex: not a bad idea. But communication might be tricky.

rsj: well, we are talking about an economy with a disfunctional monetary policy, and with a disfunctional response to that disfunctional monetary policy. Whether it gets the mix of goods right is a lesser issue. But, if some real shock did make people poorer, a functional economy would allow them to work harder to earn more income to recover their wealth.

If the existing shareholders adopt Panglossian expectations, that is one of the ways in which the debt/equity ratio matters (makes the Modigliani Miller Theorem false).

Tom. It seems to me that both wages and prices are (roughly) equally sticky. But at this level of abstraction, the distinction between wages and prices is a minor detail. It matters that one of them is sticky, but it matters less which one of them is sticky. We are just exploring a theory, to see if the alpha version can get off the ground.


"But, if some real shock did make people poorer, a functional economy would allow them to work harder to earn more income to recover their wealth."

I guess it depends on the shock and on your assumptions of people's reactions.

However, I have never seen a single firm that, in response to a decrease in sales, decides to increase hiring.

That is what you are requiring here but it seems incredibly unrealistic. Moreover, I can't think of any plausible reasoning process that would support this type of behavior.

> However, I have never seen a single firm that, in response to a decrease in sales, decides to increase hiring.

That's exactly what startups do. The entire dot-com bubble was predicated on firms flush with venture capital attempting to "capture market share" at a loss in order to monetize a bit later.

No, start ups do not do this. Start ups operate at a loss -- and even then, this is not nearly as ubiquitous as the media would have you believe -- But if sales *fall* for a start up, it will likely be shut down, as the start up is expected to *grow sales*, and can operate at a current loss because sales are at a low level but _increasing_ at a rate that justify current expenditures.

Also, start ups are irrelevant to the labor market.

Actually, looking at this again we're begging the question. In this model, firms don't choose their production quantity, they choose the price and then produce enough to meet demand. We don't have investment in inventory or capital in this simplified model.

In the face of lower presumed demand, firms can and do lower prices, even if they expect to sell the same or reduced quantity at that lower price. This applies for both a preference and AD shock, since as you say they look very similar to a firm.

The interesting thing about Panglossian expectations is not that they cause a firm to hire more, but that they may cause a firm to maintain or increase its price-point.

We did perhaps see evidence of this historically with Apple computers, which in the 90s retreated from the low-price "mac clone" market to focus on their upscale integrated systems. In the face of a preference shock (price-conscious computer purchasers were still going with PC/Windows) and a serious solvency risk, they retreated upmarket.

Why is modeling this so hard when we've already assumed part of what we're trying to get in price stickiness? i.e. that liabilities are fixed in nominal terms -- so you already have something that's sticky in your model, without explanation.

nivedita: ? Modelling *ought* to be hard, sometimes. In fact, modelling ought to be *impossible*, sometimes. Because what we are doing here is trying to see if a verbal argument hangs together, under reasonable conditions. Does it make sense? Under what conditions would it make sense?

I have a friend who ran a civil engineering business. In the depths of the recession he was out of work and eventually went bust after long months idle. Why didn't he lower the price he (his firm) was willing to work at, and work harder, so as to avoid going bust?

Because whatever work he took had to be profitable, on pain of going bust even faster. Civil engineering contracting requires large expenditures on renting machinery, buying cement, paying subcontractors, etc. If those prices do not lower, then there is no *profitable* work to be had at any price.

Note that this is a coordination problem: maybe the cement suppliers and subcontractors are out of work too, and could get some (from civil engineering firms) by lowering their prices. But they have suppliers too, and have to work profitably. You need everyone to lower their prices simultaneously so as to maintain profitability all around. Even if everyone were desperate for work, this might take time. But you only need one non-desperate agent, holding out for his accustomed price, to make the coordination problem impossible to solve in an interconnected economy.

Another example of the same phenomenon: traveling salesmen. If they are out of work, why don't they lower their wage? Well, they have to make enough to pay their travel expenses. They need hotels and gasoline to get cheaper too. Maybe hotels are underbooked, and would like to lower their prices, but they have to pay the cleaning ladies. And so on. But if the price of gasoline doesn't move much, it can screw everyone.

Note that this version of the coordination problem does not require any debt or forward-looking obligations, though that undoubtedly figures in the full explanation.

I'm not an economist so I may be missing something. But real-life experience seemed to provide a solution to Nick's initial objection independent of Steve's supply-side capacity constraint.

Heath: but see my previous post on this.

Even if each agent's profit-maximising price is 99% determined by other agents' prices, so for each agent i, the reaction function is P(i) = 0.99P + 0.01P*, where P is the average price level and P* is the price level that ensures "full employment" for all, the Nash equilibrium is still P(i)=P=P*.

Now, it's reasonable to ask: how easy is it for them to find that new Nash equilibrium, if it changes? How long will it take them all to learn it has changed, and get there?

Nick, I'm trying to say that modeling price stickiness should be hard if you're also trying to figure out an explanation for why people's liabilities tend to be fixed in nominal terms. If you assume that to start with, that's already going to feed into making prices for goods partly sticky. Suppose you assume that the price level fell by enough to offset a contraction in money supply, so you would expect real output to be unchanged. Well people with nominal debts are now poorer, and people with nominal assets are richer, but we would expect the ones with debt to cut their consumption by more than the ones with assets increase their consumption. Hence output would be lower than before, which indicates that what really happens is that both price and output fall, and price does not fall to the full-employment level.

And I think that should hang together even without having to assume that some people end up bankrupt which create additional deadweight losses etc.

Hm, I can't make that description, which sounded so obvious verbally, actually work out in a simple model if the creditors can't do anything with their income other than spend it on consumption.

nive: that's why (unfortunately) we need to model ;-)

As I think more about this, is there a special problem why agents don't negotiate prices in epsilon-increments during a recession? After all, they don't negotiate prices in epsilon-increments any other time. Businesses don't give workers regular weekly raises of a few dollars a week; I can't negotiate a few cents off the price of eggplants at the grocery store. Changes in prices are ALWAYS lumpy and they take time.

Given that, is it still a big mystery why prices in a recession are sticky and finding a new Nash equilibrium takes time?

Heath: that's basically the idea behind the "small menu cost" models. There's a fixed cost of changing price (printing new restaurant menus) independent of the size of the price change. So it's never worth it to change price by a small amount. And in the right sort of model, small costs of changing price, at the level of the individual firm, can make prices very sticky at the aggregate level. That's a coordination problem. They can't all jump their prices at once.

Many restaurants post their prices on newly written boards each day. And web sites make it easy for industrial products. And yet, prices are still sticky. Force of habit or is the real stickyness on the buyer's side?

@Jacques Someone still needs paid to rewrite the boards and update the web pages. You can change magnitudes, but you can't discount the concept (as concept).

Same on the buyer side, where I understand the correlate is called shoe-leather cost (keep having to call the restaurant (or walk by) or check the web page yet again to see if any prices have changed).

As the wikipedia articles kinda get to, it seems ultimately to be a communication cost, and communication is kinda fundamental in actual (rubber meets the road) economics.

There's also a lot of meta-communication that likely contributes to the "stickiness" phenomenon -- frequent price changes change consumers' perception/opinion of a seller, for just one example.

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad