Labour, Credit and Commodity Markets: The Partial Equilibrium Approach
Arguably, one of Keynes’s most successful methodological innovations was “macromarkets”: macroeconomics’ distinctive way of theorizing makes sense of the economy as a whole by aggregating all markets in the three “representative” markets for labour, capital and goods.
Thus equilibria with Keynesian features may be thought of as resulting from some specific failure in the working of these markets. The underlying idea is that both theory and empirics militate against the traditional strategy of using the perfect competition, auction-like models as a first approximation to reality. Two observations are called for in this respect. First, this change of focus was made possible by the many novel results on asymmetric information, contract theory, and so on, which were increasingly available by the end of the 1970s. These made it clear that there existed key assumptions (most notably perfect information), which could not be maintained as sensible approximations justified by practical considerations: for example, markets where information is imperfect work in their own way, and one had better take this into account. This in turn allowed an explicit analysis of many features, which had long been recognized as relevant, but the theory for which had not been available before: thus, for example, adverse selection on the credit market is clearly mentioned by Adam Smith but, according to Stiglitz and Weiss (1981: 394), it was their own paper which offered “the first theoretical justification of true credit rationing”.Secondly, in most instances this approach led to models where endogenous rationing is an equilibrium, thus providing theoretical support for prices not adjusting in the face of excess demand or supply in key markets: equilibrium is inconsistent with full employment. However, precisely because rationing is grounded on rational behaviour, the price rigidity one is concerned with is real (as opposed to nominal) - it is about relative prices.
This, in spite of its currently uncontroversial label, left some authors in doubt as to the Keynesian ancestry of this brand of new Keynesianism.As is well known, the basic neoclassical account of the labour market is based on perfect competition: under standard assumptions, an equilibrium (real) wage rate exists, such that the allocation of labour is “optimal”, and no involuntary unemployment occurs: no worker is unable, when willing, to work at the equilibrium wage rate. In this setup, involuntary unemployment can be observed only if the wage rate is above market clearing, and for some reason no tendency arises for it to fall. Add in that in practice this is also somewhat of an empirical statement, as in general the real wage rate does seem to be pretty slow to react to unemployment (if at all), and the case is set for asking, what theoretical reasons can be given for this downward rigidity? No straightforward answer presents itself, as a standard answer is available which is deep-rooted in elementary economics: if workers are rationed, they are off their supply curve of labour - they should be willing to accept a lower wage rate, which would benefit the firm. That is, there are obvious gains from trade to be reaped, and it is not clear why they should not materialize.
One way to introduce the NK approach is by noticing that the picture outlined above rests on an underbidding argument: gains from trade are realized by arbitrage across workers leading to full employment. Put in a nutshell, the NK argument is that the arbitrage picture of the labour market is not warranted at all: it is not obvious that gains from trade are correctly perceived, neither is it obvious that, when perceived, they are seized upon. More precisely, that the unemployed be able to exert a downward pressure on the wage rate, and that firms find it convenient to pay a lower one, are both necessary conditions which on closer scrutiny turn out to be far from trivial. Notice that they are not independent of each other: if for some reason firms are unwilling to cut down on wages, this prevents outside workers from competing with the insiders - which the latter can turn to their own advantage, whenever they have it in their power to affect the firm’s wage policy.
The NK objections to the underbidding argument are meant to be consistent with the agents’ rationality, which means that utility or profit maximization should take into account some additional constraint: usually known as the “wage setting schedule” (WSS), this prevents the real wage from clearing the market - the WSS includes the key information assumed away in the NC picture, thus giving a formal account of why the labour market is not an auction market. The two main classes of NK models, which convey these ideas, go under the name of efficiency wage (Akerlof and Yellen 1986), and insider-outsider (Linbeck and Snower 1988) models. Both deliver a WSS: in the former case, to the effect that firms may be loath to accept lower wages as this may affect productivity; in the latter, to account for the turnover costs borne by the firm, which make inside workers imperfectly substitutable with outsiders.Suppose that labour productivity increases with the real wage. Then to any employment level there corresponds a cost minimizing wage we, say, which will not be zero, as the cost-raising effect of higher wages for given productivity will be reduced by its costreducing effect via higher productivity: this is the efficiency wage. Clearly, if we is higher than the wage required by workers on their supply curve, involuntary unemployment will ensue: the potential increase in the firm’s profits given by a lower wage would be more than compensated by the higher costs due to low productivity. This is in a nutshell the logic of the efficiency wage theory, whose linchpin rests with the relationship between productivity and the real wage rate, which much of the NK literature grounded on asymmetric information: if the quality of the worker’s labour input is not observable by the firm, higher wages may be required to elicit unobservable effort on the worker’s part. In the work by Shapiro and Stiglitz (1984), perhaps the best known along these lines, equilibrium unemployment acts as a “workers’ discipline device”: for unobservable effort to be forthcoming a “high” wage is called for, while the implied unemployment deters unobservable shirking.
Suppose instead that substituting inside workers with outsiders implies some specific cost: hiring may require the firm to put up resources for training new workers; firing may entail a severance pay. This in itself will drive a wedge between the firm’s opportunity cost and the workers’ - if insiders are paid w, outsiders will be able to compete with them only by accepting (w - c). Moreover, c (the turnover cost) may be endogenous: a typical distinction is between hiring costs (in the main technology driven) and firing costs (over which insiders will have a say, more obviously so when unionised). If wage setting is somehow affected by the insiders’ own objective, turnover costs and actual wage may be so set as to price outsiders out of the market: it is enough that the resulting (w - c) is lower than their reservation wage.
Models of this kind make a case for equilibrium rationing on the labour market: they provide an equilibrium theory of involuntary unemployment, and no doubt this is why they are regarded as part of the NK research programme. However, they predict that the real wage rate will fail to clear the market: in both cases the WSS embodies constraints to utility or profit maximization, and no rational agent would care about purely nominal constraints. As a result, macro models built on these premises will deliver a vertical aggregate supply curve.
As Joseph Stiglitz (1992: 269) put it, “capital is at the heart of capitalism: it is, accordingly, not surprising that we should look to failures in the capital markets to account for... fluctuations in output and employment”. The significance of this remark lies in the underlying account of these failures, which the NK approach firmly grounds on informational considerations. Two points stand out in this respect: the idea that capital market imperfections should be seen as the rule, as they can be traced back to pervasive informational problems; and the claim that such imperfections make a case for unemployment as an equilibrium.
Credit rationing is perhaps the best-known instance.
That “a fringe of unsatisfied borrowers” was a possibility is clearly envisaged by Keynes himself in the Treatise (1930: 327), and the NK approach gave a new foundation to the idea. The starting point is again asymmetric information. Suppose there is an excess demand for credit at the going interest rate: the profitable move for the lender (a bank, say) might apparently be to raise its rate at the market clearing level; however, if investment projects are distinguished by their “riskiness”, and the latter cannot be observed by the bank (though it is known to the borrower), the positive effect on profits from such a choice can be offset by a higher probability of the borrower defaulting - in which case the profit maximizing rate will be kept below market clearing, and credit will be rationed. This may happen, as higher lending rates lead to worsening the average quality of the pool of credit applicant - as Adam Smith put it, “sober people... would not venture into the competition” (1776 [1976], hereafter WN, II.iv.15) for highly priced credit. That is, risky projects are more likely to be drawn into the market, because of limited liability: a high risk investment can bear a high interest cost, since high risk means high expected profit for the borrower, his liability being indeed limited (that is, expected losses bounded); on the other hand, asymmetric information means the lender is unable to discriminate borrowers ex ante, and so adjust its rate to riskiness. Among models based on this intuition, Stiglitz and Weiss (1981) had perhaps the greatest impact; this literature grew enormously in the 1980s (for a general assessment, see Jaffee and Stiglitz 1990).One immediate implication of this kind of result concerns policy: if credit markets are rationed, interest rate variations are a poor explanation for aggregate demand fluctuations - which does seem to be empirically the case, as the interest-elasticity of aggregate spending is typically low. Hence, monetary policy should work best when centred on managing the quantity of credit rather than the quantity of money.
This, however, is a statement about the working of monetary policy, not its effectiveness: for a theory of unemployment, a theory of aggregate supply is needed - do imperfections on the capital markets have any bearing on equilibrium output?The answer to such a question is best seen by looking at credit rationing as an instance of a more general principle: under asymmetric information the Modigliani-Miller (MM) theorem does not hold. The theorem works out a key consequence of perfect capital markets: the financial structure (for example, debt/equity ratio) of the firm is immaterial to its choices, as the cost of capital is independent of the legal form under which capital is forthcoming - investors will require the same rate of return, whatever the asset they hold. Clearly, if credit is rationed this is not the case: equity cannot be substituted with debt at no additional cost, if the latter option is constrained; but, more generally, contract forms should be seen as solutions to informational (and hence incentive) problems: thus, for instance, whether capital is raised by issuing equity or debt is not immaterial - for example, the latter entails the possibility of default, that is, a switch in the control of the firm’s assets.
One important implication of these premises is that information costs are a key component of the cost of raising capital, as different contract forms embody different solutions to different information asymmetries - internal capital will be cheapest as no contract with external investors is required, consistently with firms apparently having a strong preference for own financing. Moreover, while these diverse costs affect the firm’s choices as to its financial structure, the latter will influence its input and output choices: if (say) a negative shock hits its business, the firm’s reaction will depend on the way capital has been raised - contrary to the MM theorem, the debt- and the equity- funded companies behave differently. In short, incentive compatibility constraints are both a real cost in the allocation of capital, and a key explanatory variable to the firms’ behaviour.
All of which brings about at least two noteworthy macroeconomic consequences. First, the firm’s employment and output choices are constrained by its asset structure. Though this point is developed differently in different models, as a general principle a high value of the firm’s assets will lower its cost of raising external finance, while high (perceived) bankruptcy risk will lower output, as the firm will tend to shun risk; given that the linchpin of these models is the real cost of capital, this means that they deliver a vertical supply curve, the position of which will typically be below full employment. Secondly, as this position depends on variables like the asset composition of the economy and the value of such assets, the relevant dynamic issue of financial fragility is brought to bear on the matter (Bernanke and Gertler, 1990): for example, an unexpected drop in money supply, via its effect on prices, will typically lower the firm’s equity value, hence raise bankruptcy risk and the cost of external capital, with negative effect on output and employment - it may also trigger multiplier effects through the consumers’ permanent income, which feeds back via demand on to the firms’ asset value. It should be noticed that, though this framework does in general provide a scope for policy, the latter is different from traditional Keynesian demand management - two otherwise identical economies will react differently to external shocks depending on their asset structures, which accordingly provide a key dimension along which policy should be assessed (for example, Gertler 1988).
We conclude this section with two general remarks. First, the NK research programme is sometimes held to include also models where imperfect competition on the goods market delivers a counter-cyclical behaviour of the firms’ mark-up - the idea being that markets tend to be more competitive in the boom phases (for example, growing demand may make collusion more costly: Rotenberg and Saloner 1986), a higher elasticity of the demand for goods feeding an increase in the demand for labour. In a way, this takes up a point about old Keynesianism famously raised by Dunlop: any interpretation of the Keynesian model having firms on their demand for labour schedule should predict that output expansion requires a fall in real wages, which is not easily detected in the data. Empirically, whether on average the mark-up is anti-cyclical is, overall, controversial.
Secondly, generally speaking this brand of New Keynesianism accounts for unemployment in a framework where the aggregate supply curve is vertical - the nominal price level is immaterial. Unemployment is traced back to turnover costs or incentive compatibility constraints driving a wedge between individual optimality and maximum social welfare: while endogenous rationing entails an equilibrium real wage (or real interest rate) inconsistent with full employment, there is no room for nominal sluggishness, and hence for policies centred on managing nominal aggregates. Since real constraints are imposed on an otherwise neoclassical setup, the NK label attached to this approach did not go unchallenged: commenting on one model along these lines, Hall (1988: 263) observes that:
[the authors] write as if there were a huge gulf between their own model and the real businesscycle model, a gulf as great as the one between Keynes and the Classics. They could equally well have portrayed themselves as members of the real business-cycle school... A much more significant watershed in macroeconomics... is between the real school... and the nominal school.
This “nominal school”, which relies on a general equilibrium approach to address the issue of nominal rigidity and stresses coordination problems as the foundations of Keynesian economics, does so within a novel framework: that of general equilibrium with imperfect competition.