Snowdon & Vane Modern Macroeconomics

that a competitive economy hit by repeated technology shocks can exhibit the kind of fluctuations that are actually observed.

On the negative side, one of the problems with calibration is that it currently does not provide a method that allows one to judge between the performance of real and other (for example Keynesian) business cycle models. As Hoover (1995b) notes, ‘the calibration methodology, to date, lacks any discipline as stern as that imposed by econometric methods … Above all, it is not clear on what standards competing, but contradictory, models are to be compared and adjudicated.’ Nevertheless calibration has provided an important new contribution to the methodology of empirical macroeconomic research. While initially the calibration methodology was focused on business cycle research, more recently calibrated models have been used to investigate issues in public finance, economic growth, industry, firm and plant dynamics and questions related to the choice of economic policy (Cooley, 1997). For more detailed discussions and critiques of the calibration methodology see Kydland and Prescott (1991, 1996); Summers (1991a); Quah (1995); Hoover (1995b); Wickens (1995); Hansen and Heckman (1996); Sims (1996); Cooley (1997); Hartley et al. (1998).

6.12 Real Business Cycle Theory and the Neutrality of Money

Real business cycle theorists claim that recent research relating to the stylized facts of the business cycle support the general predictions of ‘real’ as opposed to ‘monetary’ theories of fluctuations. But, as we noted earlier, the correlation between money and output is an accepted stylized fact. How do real business cycle theories deal with the apparent causal influence of money?

Monetary neutrality is an important property of real business cycle models. In such models neutrality applies to the short run as well as the long run. In the late 1970s, leading representatives from the other major schools of thought, such as Tobin, Friedman and Lucas, all agreed that the rate of growth of the money supply has real effects on the economy and plays an important role in any explanation of output fluctuations. There was of course considerable disagreement on the nature and strength of the relationship between money and output and on the relative power of monetary and fiscal policy, but economists of all persuasions took it for granted that monetary phenomena were crucial to business cycle research. The accepted business cycle stylized fact that money and output exhibit positive correlation, with money leading output, was taken by many as strong evidence of causality running from money to output (Sims, 1972). The research of Friedman and Schwartz (1963, 1982) added further weight to the monetarist claim that monetary instability lies at the heart of real instability. However, the well-established positive association between money and aggregate output may simply indicate that the money supply is responding

to economic activity rather than the reverse. In such a situation money is endogenous and the money-to-output correlations that we observe are evidence of reverse causation; that is, expectations of future output expansion lead to current increases in the money supply. According to real business cycle theories, the demand for money expands during business expansions and elicits an accommodating response from the money supply, especially if the monetary authorities are targeting interest rates (see Barro, 1993, chap. 18). The impetus to downgrade the causal role of money was also given support from the evidence emerging from vector autoregression analysis which indicated that, once interest rates were included among the variables in the estimated system, money ceased to have strong predictive power. The contributions from Sims (1980, 1983) and Litterman and Weiss (1985) provided important evidence which advocates of the real business cycle approach point to in support of their preference for a non-monetary approach to business cycle modelling (see also Eichenbaum and Singleton, 1986).

Initially real business cycle models were constructed without monetary features. Kydland and Prescott (1982) originally set out to construct a model which included only real variables but which could then be extended to take into account nominal variables. But after building their real model Kydland and Prescott concluded that the addition of a monetary sector may not be necessary since business cycles can be explained almost entirely by real quantities (see Prescott, 1986). Although the Long and Plosser (1983) model contains no monetary sector, King and Plosser (1984) explain the historical association between money and output as reflecting an endogenous response of money to output. Building on the work of Black (1987) and Fama (1980), King and Plosser reject the orthodox monetarist interpretation of money-to- output causality. In their model, ‘monetary services are privately produced intermediate goods whose quantities rise and fall with real economic developments’. King and Plosser view the financial industry as providing a flow of accounting services that help to facilitate market transactions. By grafting a financial sector on to a general equilibrium model of production and consumption, King and Plosser show how a positive correlation between real production, credit and transaction services will arise with the timing paths in these co-movements dependent on the source of the variation in real output. Their model implies that the volume of inside money (bank deposits) will vary positively with output. Furthermore, the fact that financial services can be produced more rapidly than the final product means that an expansion of financial services is likely to occur before the expansion of output. The stock of bank deposits is therefore highly correlated with output and a leading indicator in the business cycle.

The money–output correlation noted above corresponds with the evidence presented by Friedman and Schwartz (1963) but from an entirely different

perspective. Whereas in monetarist models exogenous changes in the quantity of money play an important role in causing movements in output, King and Plosser stress the endogenous response of deposits to planned movements in output. In effect the output of the financial sector moves in line with the output of other sectors. However, by the end of the 1980s, despite the progress made by REBCT in explaining the money–output correlation, Plosser’s (1989) view was that ‘the role of money in an equilibrium theory of growth and fluctuations is not well understood and thus remains an open issue’.

Paradoxically the REBCT argument that money is endogenous is also a major proposition of the Post Keynesian school (see Kaldor, 1970a; Davidson, 1994). For example, with respect to this very issue of money-to-output causality, Joan Robinson (1971) suggested that the correlations could be explained ‘in quantity theory terms if the equation were read right-handed. Thus we might suggest that a marked rise in the level of activity is likely to be preceded by an increase in the supply of money.’ In an unholy alliance, both Post Keynesian and real business cycle theorists appear to agree with Robinson that the quantity theory equation (MV = PY) should be read in causal terms from right to left. Orthodox Keynesians have also raised the issue of timing in questioning money-to-output causality. Tobin (1970) showed how an ultraKeynesian model could be constructed where the money supply is an endogenous response to income changes. In this model changes in real economic activity are preceded by expansions of the money supply as firms borrow funds from the banking sector in order to finance their planned expansions. Tobin demonstrated that to infer from the timing evidence that changes in the money supply are causing changes in real economic activity was to fall foul of the post hoc ergo propter hoc (after this therefore because of this) fallacy. However, although Tobin used this argument to challenge what he considered to be the exaggerated claims of monetarists relating to the power of monetary forces, he certainly did not conclude that money does not matter for business fluctuations (see also Cagan, 1993).

Kydland and Prescott (1990) have questioned the whole basis of this debate on timing and causality by rejecting one of the ‘established’ stylized facts of the business cycle relating to monetary aggregates. They argue that ‘there is no evidence that either the monetary base or M1 leads the cycle although some economists still believe this monetary myth’. Clearly such claims represent a serious challenge to conventional views concerning the role of money. This ‘blasphemy’ has been rejected by Keynesian and monetarist economists alike who, as a result of real business cycle analysis, have been thrown into an alliance which would have seemed unthinkable during the intense debates that took place between Tobin and Friedman during the 1960s and early 1970s. (For a defence of the earlier Friedman and Schwartz research, see Schwartz, 1992.)

6.13Measuring Technology Shocks: The Solow Residual

If technology shocks are the primary cause of business cycles, then it is important to identify and measure the rate of technological progress. Given the structure of real business cycle models, the key parameter is the variance of the technology shock. Prescott (1986) suggests that Solow’s method of measuring this variance is an acceptable and reasonable approach. Solow’s (1957) technique was to define technological change as changes in aggregate output minus the sum of the weighted contributions of the labour and capital inputs. In short, the Solow residual measures that part of a change in aggregate output which cannot be explained by changes in the measurable quantities of capital and labour inputs. The derivation of the Solow residual can be shown as follows. The aggregate production function in equation (6.13) shows that output (Y) is dependent on the inputs of capital (K), labour (L) and the currently available technology (A) which acts as an index of total factor productivity:

Output will change if A, K or L change. One specific type of production function frequently used in empirical studies relating to growth accounting is the Cobb–Douglas production function, which is written as follows:

In equation (6.14) the exponent on the capital stock δ measures the elasticity of output with respect to capital and the exponent on the labour input (1 – δ ) measures the elasticity of output with respect to labour. The weights δ and 1 – δ measure the income shares of capital and labour, respectively (see Dornbusch et al., 2004, pp. 54–8 for a simple derivation). Since these weights sum to unity this indicates that this is a constant returns to scale production function. Hence an equal percentage increase in both factor inputs (K and L) will increase Y by the same percentage. By rearranging equation (6.14) we can represent the productivity index which we need to measure as equation (6.15):

Because there is no direct way of measuring A, it has to be estimated as a residual. Data relating to output and the capital and labour inputs are available. Estimates of δ and hence 1 – δ can be acquired from historical data. Since the growth rate of the product of the inputs will be the growth rate of A plus the growth rate of Kδ plus the growth rate of L1–δ , equation (6.15) can be

rewritten as (6.16), which is the basic growth accounting equation that has been used in numerous empirical studies of the sources of economic growth (see Denison, 1985; Maddison, 1987).

Equation (6.16) is simply the Cobb–Douglas production function written in a form representing rates of change. It shows that the growth of output (∆ Y/Y) depends on the contribution of changes in total factor productivity (∆ A/A), changes in the weighted contribution of capital (δ∆ K/K) and changes in the weighted contribution of labour (1 – δ )(∆ L/L). By writing down equation (6.15) in terms of rates of change or by rearranging equation (6.16), which amounts to the same thing, we can obtain an equation from which the growth of total factor productivity (technology change) can be estimated as a residual. This is shown in equation (6.17).

In equation (6.17) the Solow residual equals ∆ A/A. Real business cycle theorists have used estimates of the Solow residual as a measure of technological progress. Prescott’s (1986) analysis suggests that ‘the process on the percentage change in the technology process is a random walk with drift plus some serially uncorrelated measurement error’. Plosser (1989) also argues that ‘it seems acceptable to view the level of productivity as a random walk’. Figure 6.9 reproduces Plosser’s estimates for the annual growth rates of technology and output for the period 1955–85 in the USA. These findings appear to support the real business cycle view that aggregate fluctuations are induced, in the main, by technological disturbances. In a later study, Kydland and Prescott (1991) found that about 70 per cent of the variance in US output in the post-war period can be accounted for by variations in the Solow residual. We will consider criticisms of the work in this area in section 6.16 below. In particular, Keynesians offer an alternative explanation of the observed procyclical behaviour of productivity.

6.14Real Business Cycle Theory and the Stylized Facts

The rapidly expanding business cycle literature during the 1980s provoked considerable controversy and discussion with respect to the ability of different macroeconomic models to explain the ‘stylized facts’. As Danthine and Donaldson (1993) point out, the real business cycle programme ‘has forced

Source: Plosser (1989).

Figure 6.9 The annual growth rates of technology and output in the USA, 1955–85

theorists to recognise how incomplete our knowledge of business cycle phenomena actually was’, and a major achievement of this literature has been to ‘free us to reconsider what we know about the business cycle’. Research in this area has called into question much of the conventional wisdom with respect to what are the established stylized facts. Controversy also exists over which model of the business cycle best explains the agreed stylized facts. For a detailed discussion of this debate, the reader is referred to Greenwald and Stiglitz (1988), Kydland and Prescott (1990), Hoover (1991), Blackburn and Ravn (1992), Smith (1992), Zarnowitz (1992b), Danthine and Donaldson (1993); Judd and Trehan (1995); Ryan and Mullineux (1997); and Ryan (2002). Here we will briefly discuss the controversy relating to the cyclical behaviour of real wages and prices.

In both orthodox Keynesian and monetarist macroeconomic theories where aggregate demand disturbances drive the business cycle, the real wage is predicted to be countercyclical. In Keynes’s General Theory (1936, p. 17) an expansion of employment is associated with a decline in the real wage and the Keynesian models of the neoclassical synthesis era also assume that the economy is operating along the aggregate labour demand curve, so that the real wage must vary countercyclically.

Referring back to Figure 2.6 panel (b) in Chapter 2, we can see that for a given money wage W0 the real wage must vary countercyclically as aggregate demand declines and the economy moves into a recession. The fall in aggregate demand is illustrated by a shift of the AD curve from AD0 to AD1. If prices are flexible but nominal wages are rigid, the economy moves from e0 to e1 in panel (b). With a fall in the price level to P1, and nominal wages remaining at W0, the real wage increases to W0/P1 in panel (a) of Figure 2.6. At this real wage the supply of labour (Ld) exceeds the demand for labour (Lc) and involuntary unemployment of cd emerges. With the money wage fixed, a falling price level implies a countercyclical real wage.

The theories associated with Friedman’s monetarism, as well as some early new classical and new Keynesian models, also incorporate features which imply a countercyclical real wage (see Fischer, 1977; Phelps and Taylor, 1977). In Gordon’s (1993) view, apart from the big oil shocks of the 1970s, there is no systematic movement of real wages but, if anything, ‘there is slight tendency of prices to rise more than wages in booms, implying coun- ter-cyclical real wages’. However, Kydland and Prescott (1990) find that the real wage behaves in a ‘reasonably strong’ procyclical manner, a finding that is consistent with shifts of the production function. The current consensus is that the real wage is mildly procyclical, and this poses problems for both traditional monetary explanations of the business cycle and real business cycle theory (see Fischer, 1988; Brandolini, 1995; Abraham and Haltiwanger, 1995; Snowdon and Vane, 1995). If the real wage is moderately procyclical, then shocks to the production function can significantly influence employment only if the labour supply curve is highly elastic (see panel (b) of Figure 6.3). However, the empirical evidence does not offer strong support for the significant intertemporal substitution required for real business cycles to mimic the large variations in employment which characterize business cycles (see Mankiw et al., 1985; Altonji, 1986; Nickell, 1990).

While the behaviour of the real wage over the cycle has been controversial ever since Dunlop (1938) and Tarshis (1939) debated this issue with Keynes (1939a), the assumption that prices (and inflation) are generally procyclical was accepted by economists of varying persuasions. The procyclical behaviour of prices is a fundamental feature of Keynesian, monetarist and the monetary misperception version of new classical models (Lucas, 1977). Mankiw (1989) has argued that, in the absence of recognizable supply shocks, such as the OPEC oil price rises in the 1970s, the procyclical behaviour of the inflation rate is a ‘well documented fact’. Lucas (1977, 1981a) also lists the procyclical nature of prices and inflation as a basic stylized fact. In sharp contrast to these views, Kydland and Prescott (1990) show that, in the USA during the period 1954–89, ‘the price level has displayed a clear countercyclical pattern’. This leads them to the following controversial conclusion:

given money supply. Aggregate demand and supply are brought into equilibrium by a rise in the price level from P0 to P2. A favourable supply shock which shifts the AS curve from AS0 to AS1 will lead to a fall in the price level for a given money supply. The equilibrium positions a, b and c indicate that the price level will be countercyclical if real disturbances cause an aggregate supply curve to shift along a given aggregate demand curve. Referring back to panel (b) of Figure (2.6), it is clear that fluctuations brought about by shifts of the aggregate demand curve generate observations of a procyclical price level. Keynesians argue that the countercyclical behaviour of the price level following the clearly observable oil shocks of the 1970s does not present a problem for the conventional aggregate demand and supply model and that such effects had already been incorporated into their models by 1975 (see Gordon, 1975; Phelps, 1978; Blinder, 1988b). What Keynesians object to is the suggestion that the business cycle is predominantly caused by supply shocks. The consensus view that prices are sometimes procyclical and sometimes countercyclical indicates to an eclectic observer that both demand and supply shocks are important in different periods. Judd and Trehan (1995) also show that this debate is further complicated by the fact that the observed correlations between prices and output in response to various shocks reflect complex dynamic responses, and it is ‘not difficult to find plausible patterns that associate either a demand or a supply shock with either negative or positive correlations’.

6.15The Policy Implications of Real Business Cycle Theory

Before 1980, although there was considerable intellectual warfare between macroeconomic theorists, there was an underlying consensus relating to three important issues. First, economists viewed fluctuations in aggregate output as temporary deviations from some underlying trend rate of growth. An important determinant of this trend was seen to be an exogenously determined smooth rate of technological progress. Second, aggregate instability in the form of business cycles was assumed to be socially undesirable since they reduced economic welfare. Instability could and therefore should be reduced by appropriate policies. Third, monetary forces are an important factor when it comes to explaining the business cycle. Orthodox Keynesian, monetarist and new classical economists accepted all three of these pillars of conventional wisdom. Of course these same economists did not agree about how aggregate instability should be reduced. Neither was there agreement about the transmission mechanism which linked money to real output. In Keynesian and monetarist models, non-neutralities were explained by adaptive expectations and the slow adjustment of wages and prices to nominal demand shocks. In the new classical market-clearing models of the 1970s, non-neutralities

were explained as a consequence of agents having imperfect information. When it came to policy discussions about how to stabilize the economy, monetarists and new classical economists favoured a fixed (k per cent) monetary growth rate rule, whereas Keynesian economists argued in favour of discretion (see Modigliani, 1986; Tobin, 1996). The main impact of the first wave of new classical theory on policy analysis was to provide a more robust theoretical case against activism (see Kydland and Prescott, 1977). The political business cycle literature also questioned whether politicians could be trusted to use stabilization policy in order to reduce fluctuations, rather than as a means for maximizing their own interests (see Nordhaus, 1975 and Chapter 10).

During the 1980s everything changed. The work of Nelson and Plosser (1982) and Kydland and Prescott (1982) caused economists to start asking the question, ‘Is there a business cycle?’ Real business cycle theorists find the use of the term ‘business cycle’ unfortunate (Prescott, 1986) because it suggests there is a phenomenon to explain that is independent of the forces determining economic growth. Real business cycle theorists, by providing an integrated approach to growth and fluctuations, have shown that large fluctuations in output and employment over relatively short time periods are ‘what standard neoclassical theory predicts’. Indeed, it ‘would be a puzzle if the economy did not display large fluctuations in output and employment’ (Prescott, 1986). Since instability is the outcome of rational economic agents responding optimally to changes in the economic environment, observed fluctuations should not be viewed as welfare-reducing deviations from some ideal trend path of output. In a competitive theory of fluctuations the equilibria are Pareto-optimal (see Long and Plosser, 1983; Plosser, 1989). The idea that the government should in any way attempt to reduce these fluctuations is therefore anathema to real business cycle theorists. Such policies are almost certain to reduce welfare. As Prescott (1986) has argued, ‘the policy implication of this research is that costly efforts at stabilisation are likely to be counter-productive. Economic fluctuations are optimal responses to uncertainty in the rate of technological progress.’ Business cycles trace out a path of GDP that reflects random fluctuations in technology. This turns conventional thinking about economic fluctuations completely on its head. If fluctuations are Pareto-efficient responses to shocks to the production function largely resulting from technological change, then monetary factors are no longer relevant in order to explain such instability; nor can monetary policy have any real effects. Money is neutral. Since workers can decide how much they want to work, observed unemployment is always voluntary. Indeed, the observed fluctuating path of GNP is nothing more than a continuously moving equilibrium. In real business cycle theory there can be no meaning to a stated government objective such as ‘full employment’ because the economy