Friday, May 17, 2013

That Safe Asset Shortage, Continued

Regular readers of the blog will know that we have been having a contradiction with Brad DeLong and the rest of the monetarist mainstream of modern macroeconomics.

They think that demand constraints imply, by definition, an excess demand for money or "safe assets." Unemployment implies disequilibrium, for them; if everyone can achieve their desired transactions at the prevailing prices, then society's productive capacity will always be fully utilized. Whereas I think that the interdependence of income and expenditure means that all markets can clear at a range of different levels of output and employment.

What does this mean in practice? I am pretty sure that no one thinks the desire to accumulate safe assets  directly reduces demand for current goods from households and nonfinancial businesses. If a safe asset shortage is restricting demand for real goods and services, it must be via an unwillingness of banks to hold the liabilities of nonfinancial units. Somebody has to be credit constrained.

So then: What spending is more credit constrained now, than before the crisis?

It's natural to say, business investment. But in fact, nonresidential investment is recovering nicely. And as I pointed out last week, by any obvious measure credit conditions for business are exceptionally favorable. Risky business debt is trading at historically low yields, while the volume of new issues of high-risk corporate debt is more than twice what it was on the eve of the crisis. There's some evidence that credit constraints were important for businesses in the immediate post-Lehmann period, if not more recently; but even for the acute crisis period it's hard to explain the majority of the decline in business investment that way. And today, it certainly looks like the supply of business credit is higher, not lower, than before 2008.

Similarly, if a lack of safe assets has reduced intermediaries' willingness to hold household liabilities, it's hard to see it in the data. We know that interest rates are low. We know that most household deleveraging has taken place via default, as opposed to reduced borrowing. We know the applications for mortgages and new credit cards have continued to be accepted at the same rate as before the crisis. And this week's new Household Credit and Debt Report confirms that people are coming no closer than before the crisis, to exhausting their credit-card credit. Here's a graph I just made of credit card balances and limits, from the report:

Ratio of total credit card balances to total limits (blue bars) on left scale; indexes of actual and trend consumption (orange lines) on right scale. Source: New York Fed.

The blue bars show total credit card debt outstanding, divided by total credit card limits. As you can see, borrowers did significantly draw down their credit in the immediate crisis period, with balances rising from about 23% to about 28% of total credit available. This is just as one would expect in a situation where more people were pushing up against liquidity constraints. But for the past year and a half, the ratio of credit card balances to limits has been no higher than before the crisis. Yet, as the orange lines show, consumption hasn't returned to the pre-crisis trend; if anything, it continues to fall further behind. So it looks like a large number of household were pressing against their credit limit during the recession itself (as during the previous one), but not since 2011. One more reason to think that, while the financial crisis may have helped trigger the downturn, household consumption today is not being held back a lack of available credit, or a safe asset shortage.

If it's credit constraints holding back real expenditure, who or what exactly is constrained?


  1. From above... "Whereas I think that the interdependence of income and expenditure means that all markets can clear at a range of different levels of output and employment."

    Let's talk about the range where markets can clear properly in terms of monetary policy.

    I have a model that makes sense of this issue. I stayed up late last night writing it down.
    Briefly, the economy has moved to a new sub-optimal monetary zone. And the Fed is still operating in the previous monetary zone. If they were to operate in the new zone, the Fed funds rate should go up to 4% (preliminary approximation). Fed monetary policy should actually be tighter than it is. Raising the Fed rate would stabilize and regulate the credit demand. Savings accounts should be paying some interest, right?

    I lay out the model and the equations here...

    I found a way to determine a shift in the monetary zone, but there must be other ways.
    Is there a problem with a monetary zone shift? No, it's just that the US economy is more like Latin American economies all of a sudden. and those countries have positive central bank rates. Monetary policy is viable within its corresponding monetary zone. It will not be viable if it tries to behave in a non-corresponding monetary zone. And that is exactly what the Federal reserve is doing now. And that is why monetary policy has no traction.
    We could be in a liquidity trap and still operate within our monetary zone. It's not like a liquidity trap always shifts the monetary zone. In fact, my model shows that the monetary zone began shifting 10 or so years ago, much before we entered the liquidity trap.

  2. I'm sorry, but I read your post and can't understand your argument. Can you translate it into more familiar language, or tell me whose work you are building on so I have a better sense of where you are coming from?

    1. Yeah, the model for effective demand is new. I have only been publishing it since March. You will only see it on my blog. It is not built upon another work. I found the basic equation and then built it around the original concept of effective demand by Keynes.
      The basic equation and subsequent model came from data between labor share and capacity utilization.

      I saw unemployment as the factor that determined the bounded range of capacity utilization around a central tendency of labor share. I have progressed quickly into many models and applications. My blog has lots of posts now. Data fits well with the model.

      The main argument for the monetary post is that labor share determines effective demand which determines the equilibrium monetary zone upon which monetary policy operates. Labor share used to be fairly constant, and by association the effective demand limit too. So the monetary zone stayed constant. When labor share fell over the past years, the associated equilibrium zone of the Fed rate shifted along with it. It was shifting gradually until the crisis shoved it over. Now we can see the shift more clearly.

      So with humility, I offer this model as a new way to understand the dynamics of the economy and monetary policy. I am still the only one working on this effective demand model. It certainly is going to need more people hammering on it than just me to know if it is a revolutionary model or not.
      I do the best I can to develop the model and then write about it within my time constraints.
      Many times I don't understand what you write either. But I think on it... research it... sit with it. Then it comes clear. One has to do the same with my model.

  3. I'd like to make an observation about the difference between your point of view and DeLong's one (as far as I understand both) because I'm thinking about it since some time now and I think it is interesting.

    I think that there are two different "trends" of thought in economics, one of wich I would call "syncronic" (typical of right leaning economists) and the other "diacronic" (typical of left leaning economist). I think that DeLong's point of view tries to link the two through a logical error.

    a) - the syncronic point of view
    The syncronic point of view is the one that is most associated to the "invisible hand" theory and is based on this logic:
    1) stuff has an utility and a price, for example an apple has a price of 1$ and the utility of being tasty and fulfilling.
    2) people are willing to pay higher price for the things that have higher utility, and will try to produce stuff that is more profitable. In other words utility determines potential demand while price determines supply.
    3) the marginal utility of stuff falls when there is a lot of it.
    4) this lead to an equilibrium in the quantity of the stuff produced that maximizes total utility. It is important that the price system is just a way to calculate the correct relative quantities of stuff, the system is supposed to reach an equilibrium through an adjustment of quantities.

    In other words, the syncronic point of view deals on the question "should we produces apples or oranges? how many?" (also known as the soviet calculation problem). There is only one equilibrium that is determined by the relative utility of apples vs. oranges and their cost of production.
    I call this point of view syncronic because it deals on the equilibrium of different stuff in the same moment.
    People who apply this syncronic point of view can explain things like recessions or depressions only as a moment when, for some reason, the optimal equilibrium can't be reached (e.g. the GFC is caused by the fact that the USA built too many hoses but the market can't reach the true equilibrium because the government subsidizes housing).

    An important aspect of the syncronic point of view is that it is often overlooked by its proponents is that it is based on the concept of utility. Only consumption goods have utility, while a lot of other intermediate goods have price (for example a car has both a price and utility, while a car factory only has price, but no utility in itself). As a consequence all goods (real or financial) that have no utility in themselves derive their price only from the utility of the consumption goods that they can produce or purchase.
    In other words, in the syncronic point of view the idea that people "want" money (or financial assets, or other capital goods) simply make no sense, people can only "want" consumption goods (in the present or in the future) and pursue money to buy those consumption goods.


  4. [...continues]

    b) the diacronic point of view
    is the point of view that I associate with theories about recessions, crises etc., like Keynes or Marx, and I think is also the point of view you usually adopt.
    This point of view frames the economy in terms of cycles of production: for example businesses today produce X "stuff" and pay Y wages. Is Y enough to pay for X and thus substain this level of production?
    The diacronic point of view usually doesn't think about "utility" and the correct proportion of different kinds of stuff produced, and just assumes that "the market" is doing its job of organizing production.
    On the other hand the diacronic point of view usually cares a lot about "wealth" (in a sense that is closer to "price" than to "utility") and the evolution and accumulation of wealth per se.
    In the diacronic point of view we can speak of different equilibria in this sense: we can think of various level of sustained production and consumption, each of wich is a possible equilibrium, and the system is in disequilibrium only when production doesn't match consumption, which causes the sistem to change equilibrium level.
    Usually levels with higer production and lower unemployment are assumed to be "better".

    However, the concept of "equilibrium" used in the diacronic point of view is completely different from the concept of "equilibrium" used in the syncronic point of view; in fact the two "schools" use the same words to mean completely different things, so that it is not correct to say that they have different ideas about equilibrium/a, but just that they speak of different things in different contexts.

    c) The DeLong/New Keynesian point of view
    The point of view that usually DeLong adopts is similar to the one that you presented as "New Keynesian" in a previous post.
    The point is that those economists want a framework that links both the syncronic and the diacronic povs.
    But, they have a problem because the syncronic pov speaks of choices about utility in a set of contemporaneous options, while the diacronic pov speaks of changes of wealth in different iterations of the production cycle, two completely unrelated levels of analysis. So they try to squeeze the diacronic pov inside the syncronic pov:

    - The New Keynesians think in term of "lifetime incomes", that is the same of saying that everithing happens contemporaneously, so that choices about savings becomes choices about consumption in different point of time (though the choice happens simultaneously). This leads also to some weird ideas like the strong version of the efficient market hypothesis that, in pratice, means that the market is able to optimize future utility.

    - DeLong imho doesn't believe in such extreme ideas, but doesn't get the point that the syncronic pov thinks in terms of utility and only consumption goods have utility. So when he says that the market "wants" more safe assets he means it in the same sense that one would say that the market "wants" more shoes or more apples. This imho is a logical error, but the syncronic pov does not have a way to deal with wealth per se, but only with wealth as a way to purchase consumption goods ("utility"), and as a consequence the idea that people "want" safe assets as an end in itself is a cntradiction in terms.

  5. RL-

    This is absolutely right, spot-on. Altho the language is a little nonstandard. Fuller response soon, I hope.

  6. If it's credit constraints holding back real expenditure, who or what exactly is constrained?

    A babushka doll of answers!

    Little doll: Before the crisis, 15% of households had FICO scores under 600, but now 25% do. The households with bad credit are credit constrained.

    Bigger doll: The loss of equity in your home constrains your ability to HELOC. Before the crisis, net heloc withdrawals were about 9% of disposable income, and since then have declined to -3% or so:

    Bigger doll: The previous doll was just a special case of the "less safe assets" argument, where houses were considered safe assets that did not lose money.

    Bigger doll: Increased uncertainty of future labor income will make you less likely to borrow, as will a history of prolonged unemployment. It doesn't really matter whether the credit constraint is imposed on you by the lender or self-imposed -- in both cases, the operative mechanism is fear that the loan will not be repaid. Your discounted labor income is your biggest asset, and is now less safe.

    Biggest doll: Credit constraints are pretty much fundamental to everything else. Banks can lose huge amounts of money but not go under because they are able to roll over loans and mark to model, so they live in the macro type view of lending. But the non-financial sector is subject to pro-cyclical credit constraints. The crudest example would be debt to income limits, so that a decline in income reduces the amount you can borrow which, in aggregate, may lead to a further decline in income.

    1. Credit constraints are pretty much fundamental to everything else.

      > “Interest rates are not a significant factor in our decision-making on investment because interest expense is only a small proportion of our cost base,” says Ulf Quellmann, global head of Treasury at Rio Tinto, a mining group.

      Are you sure they're so fundamental?

    2. Yes.

      That one particular firm doesn't care about one particular cost of borrowing is not important. The firm certainly has an overall cost of capital that it must meet, and if does not meet it, heads will roll -- in this case, CEO heads followed by employee layoffs. Rio Tinto's share price fell from the $60 range to the $40 range because it failed to earn the return demanded. The CEO was fired and we have a new plan to woo investors -- austerity!

      New BHP, Rio Tinto bosses woo investors with austerity talk

      New bosses at two of the world's largest mining companies, BHP Billiton and Rio Tinto, wooed investors on Tuesday with promises to slash billions of dollars of spending and press ahead with asset sales, boosting returns.

      Mining companies have come under pressure from investors for splashing out on pricey projects during the industry's boom years - at the expense of shareholder returns - while costs spiralled out of control."

    3. @RJS: "Credit constraints are pretty much fundamental to everything else."

      @Isomorphismes: "Are you sure they're so fundamental?"

      @rjs: "Yes."

      Worth noting: In the National Federation of Independent Businesses surveys since 1986, business owners have consistently put Finance and Interest Rates *dead last* on their list of business constraints. This even in the thick of the recent so-called "credit crisis."

      In 2009, German manufacturers reported that credit was easier to get than it was in 2003, 2004, and 2005. Was any one talking about credit crunch back then?

      In 2012, only 20% of German businesses were reporting that credit was restrictive -- down from 60% in 2003.

      Nonresidential Investment/GDP hit an all-time high in...wait for it...1980.

      It hit a 48-year low in 2010/2011, and is only slightly recovered since -- all while interest rates are at unheard-of, historic lows. (cf. Josh's point about junk-bond rates)

      All of that is pretty hard to explain if interest rates are such a dominant factor in businesses' investment decisions...

    4. I'm not sure what you think this shows.

      Credit constraints exist because of incomplete markets. A worker is not able to hedge their labor income just as a firm is not able to hedge its earnings. Both sides face volatile income streams but are expected to make payments on the liability side that are independent of what the income stream is.

      The constraint is that they cannot find a counter party to guarantee their income. That concern is shared by both the party in question and any future creditor (including shareholders), but you cannot measure this reluctance to expand balance sheets purely by looking at the returns demanded by creditors because that only captures part of the story; the other part being those who would like to borrow but choose not to even try as a result of the inability to find an income hedging counterparty.

    5. @RJS: If you haven't read it already, you'd probably really like Shiller's "The New Financial Order." Proposes a whole lot of financial options addressing what you describe.

    6. I like Schiller, and I should read this. Also, Miles Kimball has proposed national credit cards to be given to the population. I believe that welfare would be improved if we gave every W2 earner received a credit card with

      1) a large credit limit -- e.g. 3x average annual income over the last few years
      2) low interest rates -- the card should charge only FedFunds with no late fees ever (risk free rollover)

      3) Once the limit is reached, you would elect a repayment plan which consisted of payroll deductions, managed by the IRS

      I think this would greatly reduced the fear (and sting) of unemployment. People would no longer need to worry about losing their homes, for example, if they lost their job for a year. During a recession, the rates would be lowered, etc. People could also borrow on the same terms as banks, and labor income would be viewed as more valuable.

  7. Unemployment implies disequilibrium, for them,

    Yeah, this sounds pretty silly in my opinion. It's leading toward words like "structural unemployment" and "out of the labour force" which are supposed to have meaning but are actually a Gordian knot of parries and thrusts and probably denote the same thing in the end. In a discipline where actions, not words.

    Just one of many problems with the conception you're arguing against.

    if everyone can achieve their desired transactions at the prevailing prices, then society's productive capacity will always be fully utilized.

    What about the fact that people can work at different jobs? What about geography? Etc.

  8. Credit constraints are pretty much fundamental to everything else.

    Of course you are correct, rsj. The final question of the post is poorly phrased. What I mean was something more like:

    What specific expenditure is more constrained by a lack of credit today compared with 2007, due to a financial intermediaries' unwillingness to hold risky assets and excess demand for safe ones?

    1. Helocs.

      And in general, people are more cautious about borrowing to buy a house and/or sinking a lot of money into renovations.

  9. rsj,

    Yes, no doubt people are more cautious about incurring liabilities secured by housing. That's not a sign of an economy-wide shortage of safe assets, though, is it?

    1. I would say that there is a shortage of marketable or liquid assets, as labor income is a non-marketable asset.

      If you need to pay your bills and are unemployed, you would like some marketable assets to sell. Same for retirement. There was a one-two punch of households having their savings reduced via house price decline just as their labor income risk increased.

      To second order, safe assets are more desired now because you would like them to be uncorrelated with the business cycle.

      To first order, a big number in your brokerage account is pretty good, too.

      More important than the level, I would say that the problem is with distribution of market assets.

      I think that even if you increase incomes, people will primarily save up until they reach their target level of marketable assets so that they can sleep easier at night. Alternately, you can provide income guarantees that will make the labor income be more valuable. For those nearing retirement, an increase in SS benefits would help.

      I don't think that you can answer this question by looking at spreads or interest rates. You might be able to get more information by looking at household expenditure surveys.

  10. Rsj,

    I'm glad you wrote this, since it's exactly the position I'm arguing against, and one wants to be sure one is arguing against something smart people actually believe.

    Let's take a step back. We're in a world where on the one hand there are hungry, unemployed people, and on the other there are orchards where apples are rotting on the ground. It would seem there's a welfare-improving trade where the hungry people sell their labor to the orchard owners in exchange for the apples they harvest. Why isn't it happening?

    One option is the RBC or New Classical answer, that this situation is impossible. If people aren't working in the orchard, that means the apples they would harvest are less valuable than their time. Nobody here believes that, obviously.

    Two is the monetarist answer. People want to hold "money." Since there is a fixed supply, and its price relative to other goods is also fixed, when people try to shift expenditure from consumption to increasing their money holdings, the result is just that incomes fall until the marginal value of an additional unit of consumption equals the marginal value of holding an additional unit of money. In the more strictly monetarist version of this story, demand for money is directly related to current income; in the Keynesian version, demand for money shows up first in the market for existing asset stocks.

    That brings us to the safe assets or credit constraints version. Here the welfare-improving trade doesn't happen because at least one of the participants would have to initially borrow for their side of the transaction. That means someone has to be willing to hold their liabilities. More specifically, the story here is that in principle, these liabilities have positive present value, but holding them involves bearing risk, which is only acceptable if the lender also holds sufficient safe assets.

    Both these stories assume that people know there is a welfare-improving trade. A fourth answer is that people don't have knowledge the whole universe of potential trades, only local knowledge. (I know you're sympathetic to this idea.) I.e., there's no auctioneer. In the safe-assets story, the orchard owner knows that it would be profitable to hire more workers, and the unemployed person knows they would be better off consuming more apples in the present period, they just can't get credit to do so. Whereas, with only local knowledge, the orchard owner may think that the current state of demand for apples will continue indefinitely so there is no point in producing more; and similarly, the unemployed person may think that their lifetime income is now low so their low current consumption is as it should be. Call this the pure coordination-failure story.

    (It also may be that credit constraints are real, but don't depend on the stock of safe assets. Maybe debt contracts aren't enforceable, so no one will lend to the unemployed person no matter how many safe assets the lenders hold.)

  11. (continued)

    It seems to me that stories two and three have several important implications compared with story four. First, they imply that only the aggregate stock of money or safe assets matter. It doesn't make any difference where they are injected into the system. So in case 3, the lender who sees a positive value loan they could make *if* they held more safe assets will bid new safe assets away from anyone else who holds them and can't use them that way. Conversely, the current level of activity doesn't matter. For each quantity of safe assets or money, there is a unique associated level of real activity. Finally, a fall in the price of current goods and factor services, *if* it could be achieved, would stimulate demand just as much as raising the stock of money/safe assets.

    Whereas in story 4, the coordination-failure story, a change in the mix of assets won't have any effect, and the effect of an increase in the total stock of assets will depend on who specifically gets the new wealth. And there is not a unique mapping from the asset stock to the level of activity -- it takes more favorable credit conditions to increase current expenditure among people who think their lifetime incomes are low, than to maintain it among people who think their lifetime incomes are high.

    And I do think the credit constraints version is testable by looking at spreads. If the reason the orchard is employing fewer people than it used to, is because it can no longer find anyone who will hold the liabilities it would need to issue to hire people, then credit markets should show some sign of reduced demand for orchard liabilities.

  12. JW,

    To take the example of the orchard, the owner of the orchard, seeing less demand for his apples, assumes that he will earn less income. The credit markets will too, so the enterprise value of the orchard shrinks and equity must absorb that. This may make it more difficult for the orchard owner to borrow and he has to pay higher rates to do so; before he had $80 of equity and $20 of debt for an enterprise value of $100, corresponding to making profits of $10 per period at some (fixed) discount. Now, he has $20 of equity and $20 of debt, corresponding to an expectation of only $4 of profits at the same discount. The orchard owner is now a riskier business.

    But the way that we measure spreads is to normalize risk and look at the yield over time. BAA-AAA spreads will not change.

    Now it does not matter whether, at the new 20% equity/debt mix corresponding to depressed sales, the orchard owner chooses not to borrow or goes ahead and borrows but the markets charge higher interest rates or allow him to borrow less (after all, he can only borrow $20 now, because borrowing doesn't increase enterprise value).

  13. .... What does matter is that the orchard owner would borrow more if his mix was 80/20, not 20/20.

    Looking at the workers in the orchard -- they now believe that their wages are at risk, so they will value the NPV of their labor income much lower. Say from $50 to $20. Now, they too will have more savings demands and will borrow less. Many will defer buying a house, or making a residential investment, or buying a car/durable good.

    Yes, this is a coordination problem, but that is what it means to have incomplete markets. If we had complete markets in all things, then the market price would serve to coordinate everything. In particular, the orchard owner could sell both present and future apples, and the workers in the orchard could sell their present and future labor, buying present and future apples, and only when all of these markets clear would the economy begin.

    That we cannot do that means that we risk coordination failure. I don't see a difference between the coordination failure explanation and the incomplete markets explanation.

    I focus on credit constraints because I believe these are the most important missing markets -- e.g. inability to hedge labor income and business revenue causes excess conservatism and prevents investment.

    If the unemployed worker *could*, he would sell a type of debt contract in which he can buy apples now and pay back the loan when he is hired again, being able to roll over the loan at the risk-free rate during the interval of unemployment with no fees. If he did that, the owner would sell more apples and hire the worker back. The owner's equity would rise, and he would have enough to borrow more to invest, etc.

    That is why I agree with Miles Kimball and think that national credit cards charging only FedFunds as interest with generous repayment terms would be a great idea. I understand that it is just one idea.

    Another idea is to have the government buy the apples from the orchard owner, taking out debt, and then paying the risk-free rate in future periods on the debt. Those payments would go to the orchard owners, not the workers, so effectively the workers are still transferring real resources to the orchard owner, except the government has an allocation problem as to who to give the apples to. Giving the workers credit cards that effectively empower them to borrow for themselves at the risk free rate would be better in this regard. This is not a substitute for increased provision of public goods during recessions, nor for unemployment insurance.

    There can be many things wrong with the economy, with many approaches needed.

  14. @rsj: reduced private sector (non-bank) spending could a priori be caused by reduced credit demand or supply. But you seem fixated, like monetarists, on credit supply constraints, to the extent of barely acknowledging demand issues (you go so far as to suggest that lower demand for credit is probably caused by would-be borrowers not even trying to borrow because of credit constraints).

    What you identify as "demand for marketable securities", I would interpret as "desire to save".

    It's not that people look at their portfolio of assets and suddenly want their assets to be more liquid. A minimum of liquidity is important, sure, but in a balance sheet recession people just want higher equity (/ lower net debt) overall.

    In Richard Koo's vocabulary, you and monetarists are stuck in a "yang" paradigm where credit constraints are the limiting factor, instead of acknowledging that we are now in a "yin" world where non-bank actors mainly want to boost their balance sheets and saving.

    What is your response to this?

  15. To me, the critical thing is a strong, reasonably stable link from current income to current expenditure. (This includes a model of investment where current sales and/or profit flows are important.) This link can be motivated either by expectations of the future that are formed by extrapolating current conditions; or by ubiquitous credit constraints that limit most units' borrowing to some multiple of their current income. I think both of these are true facts about the world, so it doesn't really matter to me in this context which is more important in maintaining the income-expenditure link. The link itself is the key thing. (As it was for Keynes- arguably the consumption function was the key innovation between the Treatise and the General Theory.)

    So I have no problem with credit constraints as a central part of the story. What I do have a problem with is the idea that any time you see persistently depressed expenditure, it must be (1) mainly or entirely due to a lower credit supply, independent of borrowers' incomes, which (2) is in turn due to changes in the proportion of safe or liquid assets in the portfolios of lenders. I don't think this idea is always wrong -- I think it plays a central role in initiating many, even most, downturns. But I do not think that it explains the persistence of a depressed state once it is established. In particular, I do not think the US at the present moment can be described as experiencing shortage of safe assets or risk-bearing capacity, and I do not see any evidence that **similarly credit-worthy borrowers** have more difficult borrowing than they did in past periods of relatively full employment.

    I think we need the income-expenditure link both to talk about policy in a sensible way, and to be able to think of the capitalist economy as a system with dynamics in historical time. For instance, you often hear metaphors like the economy having a "stall speed," or being like a car at the bottom of an icy slope -- with enough speed, it will make it to the top and then be stable there, but with moderate speed it will end up right back where it is now. These kinds of multiple-equilibria situations come very naturally from income-expenditure models, but are much harder to justify in terms of varying credit constraints. In the safe assets story, in particular, there is a unique level of output associated with each aggregate asset mix, so it makes no sense to talk about stuff like stall speed.

  16. J.W.,

    I agree that there is an income-expenditure correlation. But I think this is descriptive, not explanatory. So you posit that the economy is divided into two groups, one which spends a proportion of its income and one which spends "autonomously".

    Why doesn't the second group spend autonomously? Why are they modelled with a cash-in-advance constraint whereas the other group is not?

    I posit that you can explain this difference by assuming, for example, that one group has primarily labor income which is unsuitable as collateral for loans, and so is subject to a cash-in-advance constraint. Of course, not entirely so, because individual households do spend in excess of their income on occasion.

    Moreover this second group -- the primarily autonomous group -- why is it autonomous, and to what degree is it autonomous?

    It chooses to spend because it expects a future income stream from current investment, and the value of that income stream, discounted by some rate, is the borrowing capacity of that group.

    The second group can refuse to invest either because the discount rate is too high, or because the expected future income stream is too low.

    But then why have two separate frameworks for the two groups?

    Why not assume that both groups are making the same trade-offs, except that the assets of one group (household labor income) is less suitable as collateral versus the assets of the second group (firms). Firms can be liquidated to their creditors whereas households cannot. Firms can promise a share of future income in a more legally binding way than households. Firms provided detailed accounting statements to creditors that households cannot (e.g. households do not have in house accountants that are somewhat independent of management), nor do they publish quarterly financial statements audited by third parties.

    And this framework explains other things as well -- in a recession, it is not just those who lose income who spend less. Everyone spends less. Even households that are still employed. Even households whose income has gone up. Why? Because all households perceive their labor income as more at risk, and therefore discount it more.

    A recession reduces only some people's incomes, but it makes everyone poorer, in terms of their present wealth.

    A pure income-expenditure model does not take account of this, nor of precautionary buffer stock savings demands more generally.

    So while I agree that economies tend to be characterized by an income-expenditure relationship, I would propose that we not use this as an assumption or constraint on behavior, but model behavior in some other way and see if we get an income-expenditure link as a result of some more fundamental processes. These processes would be changing the discount rate applied to different types of income, changing expected future income, and in the case of the current recession, changing the collateral value of the single biggest non-labor asset that households have, namely housing wealth.

  17. while I agree that economies tend to be characterized by an income-expenditure relationship, I would propose that we not use this as an assumption or constraint on behavior, but model behavior in some other way and see if we get an income-expenditure link as a result of some more fundamental processes.

    Sure, you and every other economist. This is all the profession has been doing for the past 30 years.

    Look: There is not a "true" model of the economy out there. We use models as tools, to think carefully about particular sets of logical relationships. It can be interesting to think about what features of credit markets might explain a regular relationship between current expenditure and current income. It can ALSO be interesting to say, given such a relationship, what kind of dynamics might we see? They're two different questions, both potentially interesting.

    But there's this weird, really almost compulsive thing in the economics profession where if you try to ask the second sort of question, people immediately start shouting that you cannot even ask it until you give them a satisfying answer to the first question.

    Yes, your project is interesting. But that does not give you any right to dismiss the DIFFERENT project that starts from aggregate relationships.

    1. Josh,
      Is, to some extent, the issue here "reductionism"? (aka, the desire for microfoundations). Reductionism does not even always serve physics well. A (social) science that has a very limited access to controlled experiments probably should not be obsessed with reducing.

  18. Is, to some extent, the issue here "reductionism"? (aka, the desire for microfoundations)

    Yes, I guess that is what I was saying in my last comment. But in general, I've come around to the view that we should NOT be arguing against reductionism in general. The question is what kind of reductionism. The real problem is that the term "microfoundations" is used in two very different senses in economics, and a great deal of confusion in discussions in methodology comes from slippage between them.

    In sense 1, "microfoundations" means that if we want to postulate a relationship between aggregates, it must in principle be compatible with some logical and empirically defensible account of the behavior of the individual units that make up those aggregates. In sense 2, microfoundations means that all economic models must be formulated epxlicitly in terms of optimization by some agent(s) with model-consistent expectations. (Aka "rational expectations", but note this is not that the meaning of rational here is not the same as in ordinary language.) In discussions with the general public, economists tend to talk about the importance of microfoundations in sense 1, while when it comes to the methodological standards enforced within the profession, it's almost always meant in sense 2. This is a big part of what it means to "think like an economist," but it's almost never spelled out. You have to spend some time with economists before you realize that when they say a story is or is not microfounded, what they actually are talking about is whether it is formalized as the maximization under constraints of some quantity, given the true expected probability distribution of all future events.

    So I think when critics of mainstream economics -- including me at times, like in the comment above -- argue against reductionism in general, we're giving modern macro too much credit, and missing the real problems. I would have no problem at all with the case for better microfoundations if that meant paying more attention to how real businesses describe their investment decisions, or more use of survey data, or closer attention to the actual patterns of household expenditure, or anything like that. But instead there's this bait and switch, where the case is made for microfoundations in sense 1, and then the actual requirement is for microfoundations in sense 2.

    (I am not so enthusiastic about behavioral economics -- I think useful microfoundations need to deal with the concrete institutions of modern capitalism, and not human behavior in general. But even behavioral econ would be a lot better than what we get.)

    Note that it's perfectly possible for a model to satisfy the requirement for sense-2 microfoundations and lack them in sense 1, in fact this is common. A growth model, say, is considered adequately microfounded if you can show the path followed by GDP would be optimal for a representative agent, even if you have no story for why the behavior of a large number of firms and households should aggregate up to that agent.

    1. So, just to give some concrete examples, consider a Minskyan story like "the longer it has been since the last crisis, the less worried people are about a future crisis, and the more willing they become to take risks"; or, Keynes story about the bond market, "In general bnd prices representat a balance between 'bull speculators' who expect interest rates to fall and 'bear speculators' who expect interest rates to rise." Both of these are perfectly microfounded in my sense 1, they are coherent stories of individual behavior. But they are not and cannot be microfounded in the formal economics sense, sense 2, because they include statements about the "expectations" in people's heads that are not simply the value of the expectations operator on the true probability distribution function.

    2. That is not true. Microfoundations is not the same as rational expectations. You can have learning. You can assume that there is a continuum of agents, each with a different estimated distribution, etc.

    3. Josh, very clear explanation and nice examples. From this (and frankly from rsj's arguments) the issue also appears the type of rationality in the mainstream models (which notoriously get trashed most other places in the social sciences).

    4. Microfoundations is not the same as rational expectations

      In principle, yes, they are different. In actual usage by most economists, I think they are the same. Do you disagree?

    5. I think that, in practice, people want easy math. I don't think the problem is principled opposition to models with more realistic descriptions of the human animal, but rather unprincipled acceptance of models with unrealistic behavioral assumptions.

  19. "(I am not so enthusiastic about behavioral economics -- I think useful microfoundations need to deal with the concrete institutions of modern capitalism, and not human behavior in general. But even behavioral econ would be a lot better than what we get.) "

    Yes. The promise of agent-based simulations -- with all of the necessary assumptions about heterogeneous and mutually modifying reaction functions (Type 1-style) -- seems to be far off. Sector-based simulations (i.e. Keen's Minsky modeling) require far fewer and less complex encoding of reaction functions. Even there, though, developing and encoding those reaction functions, and testing the resulting models through back-bearings (what else do we have?) is going to be a very long project, I think.

  20. @RSJ There seems to be some of what Rob Kuttner calls the is-ought confusion with regard to incomplete markets.

    By incomplete market do you mean markets of any kind or just competitive markets? Reality seems to serve up a rich menu of potential market failures, asymmetries, moral hazard, rent-seeking and so on. And the constellation of possible failures varies according to the current political and institutional realities.

    The reason why in principle we should not treat all individuals as mini-portfolio managers is because that model of rationality seems woefully inadequate to account for reactions to all varieties of market failure out there (especially information asymmetry), for starters.

    And frankly, doesn't the tendency toward rent-seeking blow up all the mainstream models.

    1. Incomplete markets is not about rent-seeking or different types of competition, although they can be viewed as contributing to these