Real Business Cycles: A New Keynesian Perspective

Mankiw states at the outset that he is “not an advocate” of real business cycle theory, and that “real business cycle theory does not provide an empirically plausible explanation of economic fluctuations,” singling out “its reliance on large technological disturbances as the primary source of economic fluctuations and its reliance on the intertemporal substitution of leisure to explain changes in employment” as its “fundamental weaknesses” (Introduction). He adds a normative warning: because RBC theory “trivializes the social cost of observed fluctuations,” it is “potentially dangerous” if policy advisers use it to argue that stabilization policy is unnecessary.

Mankiw explains that once money is appended to a Walrasian general-equilibrium model via a simple money-demand function, the model exhibits the “classical dichotomy” (Patinkin 1956): real variables (employment, output, relative prices) are determined entirely within the real, Walrasian system, while nominal variables (the price level, nominal wage, nominal interest rate) are pinned down afterward by the money market, making money “largely irrelevant” to real outcomes. He argues every macroeconomist must either destroy this dichotomy or live with it: Keynesian theory destroys it by assuming wages and prices do not clear markets instantaneously (motivated by long-term labor contracts and sticky product prices), while the early 1970s new classical models (Lucas 1972, 1973) tried to destroy it without abandoning continuous market clearing, via an imperfect-information/misperceptions mechanism that “attracted relatively few adherents in the 1980s” because, in Mankiw’s view, “it is hard to believe that confusion about the price level is sufficiently great to generate the large changes in quantities observed over the business cycle.” Real business cycle theory, by contrast, is distinctive for fully embracing the classical dichotomy and “the complete irrelevance of monetary policy.”

Mankiw identifies as “an obvious but important fact” that consumption and leisure move in opposite directions over the business cycle (consumption falls and leisure rises in recessions), which is “potentially problematic” for RBC theory because, since both are normal goods, most demand-side disturbances (e.g., a temporary increase in government purchases) would be expected to move them together, not apart. The only way RBC theory can reconcile this is if the real wage – the relative price of leisure – is procyclical, which Mankiw calls “a crucial implication of real business cycle theory.” He argues that if the aggregate production function were stable, diminishing marginal returns to the (cyclically low) labor input in a recession would imply a countercyclical, not procyclical, real wage; RBC theorists therefore require large fluctuations in the available technology itself, so that the real wage (and the marginal product of labor) are low in recessions for technological, not merely compositional, reasons.

Following Prescott (1986), Mankiw presents his own calculation of the Solow residual (total factor productivity) for the U.S. private economy excluding agriculture and housing services since 1948, and confirms “substantial fluctuations in measured total factor productivity” – for example, a 3.5% rise in 1984 against a 3.4% fall in 1982 – and that “in every year in which output fell, productivity also fell.” He argues, however, that the Solow residual need not reflect genuine exogenous technology shocks: the standard alternative explanation is that measured productivity falls in recessions because firms engage in labor hoarding (“off the production function” behavior), retaining underutilized workers who are worked harder once a boom returns workers to full effort. As a striking illustration that the Solow residual is not a reliable short-horizon technology measure, Mankiw notes that measured total factor productivity grew at 7.6% per year from 1939-1944 during the demand-driven World War II buildup – far above the 5.2% record for any peacetime year (1950) – which he reads as evidence the Solow residual reflects demand-driven labor hoarding/effort variation rather than true technological change over short horizons.

Mankiw argues that if society had suffered the large adverse technological shocks RBC theory requires, “we would be aware of it,” pointing to the 1970s OPEC oil-price increases as a genuine real shock that was “easily identifiable and much discussed” – yet while the Solow residual indeed shows large negative values in 1974 and 1979 (consistent with the oil shocks), “the five other recessions in the post-war period also exhibit large negative Solow residuals” without any comparably identifiable adverse technological event, which he takes as evidence those recessions “cannot be easily attributed to exogenous real shocks.”

Mankiw observes that over the business cycle, employment varies substantially while the real wage and real interest rate – the variables that are supposed to drive labor-supply decisions in RBC models – “vary only slightly,” so that mimicking the observed pattern requires individuals to be “very willing to reallocate leisure over time” in response to small relative-price movements. He argues this is empirically implausible on two grounds: econometric studies of individual labor supply (Altonji 1985; Ball 1985) find that “expected changes in the real wage lead to only small changes in hours worked,” and, more informally, that “the real interest rate is simply not a significant consideration when individuals decide to leave their jobs or to accept new employment” – concluding that “the intertemporal substitutability of leisure is very likely far too weak to get real business cycle models to work.”

Mankiw notes that multi-sector RBC models (Long and Plosser 1983) can generate an aggregate business cycle from independent sectoral technology shocks through wealth effects on demand, but only if “the number of independent sectoral shocks cannot be too great,” since with many independent shocks and mobile labor the law of large numbers would average them to zero – so these models “require that there be only a few sectors… subject to large technological disturbances” and thus “suffer from the same weaknesses” as single-sector RBC theory. He separately evaluates Lilien’s (1982) sectoral-reallocation theory, under which recessions reflect an unusually high need for intersectoral labor adjustment, and judges that “the available evidence appears not to support this intersectoral story”: if workers were voluntarily unemployed while searching for jobs in other sectors, high unemployment should coincide with high job vacancies, but observed fluctuations show the opposite pattern (high unemployment with low help-wanted advertising, per Abraham and Katz 1986), and measured intersectoral worker mobility is in fact strongly procyclical (Murphy and Topel 1987) – the reverse of what the theory predicts.

Mankiw argues it is “ironic that real business cycle theory arose in the wake of Paul Volcker’s disinflation,” since Volcker’s announced slowdown in money growth was followed by one of the deepest postwar recessions and an eventual fall in inflation – a sequence “easy to explain” under Keynesian sticky-price theory but “less easy to explain” under RBC theory (citing Romer and Romer’s 1989 finding that output typically falls after the Fed makes an explicit anti-inflation decision, which they read as evidence against RBC theory). He presents King and Plosser’s (1984) response – that standard money measures like M1 are mostly “inside money” created endogenously by the banking sector as one more sector’s “output,” so procyclical money reflects money responding to, not causing, output – but argues this account, “while [it] can explain the procyclical behavior of money… cannot explain the procyclical behavior of prices,” since King and Plosser’s own model implies rising demand for real outside money in a boom should make the price level countercyclical, when in fact (absent identifiable real shocks like the OPEC increases) inflation is well documented to rise in booms and fall in recessions – the classic, empirically robust short-run Phillips curve correlation that “has thus been left without an explanation by real business cycle theorists.”

Mankiw frames the choice as a tradeoff between “internal consistency” (a parsimonious theory with no ad hoc axioms) and “external consistency” (a theory that fits the facts): real business cycle theory “extends the Walrasian paradigm… and provides a unified explanation for economic growth and economic fluctuations,” while new Keynesian theory “relies on nominal rigidities that are observed but only little understood,” and in some versions even questions the axiom of fully rational, optimizing individuals – “the ultimate abandonment of internal consistency” by the standards of economic theorizing. He explicitly casts this as an unresolved and continuing tension within the broader scientific enterprise of macroeconomics, not a dispute with an obvious resolution on theoretical grounds alone.

Mankiw predicts that “real business cycle advocates will not manage to produce convincing evidence that there are substantial shocks to technology and that leisure is highly substitutable over time,” and that “[w]ithout such evidence, their theories will be judged as not persuasive,” while new Keynesians “have made substantial progress in recent years toward providing rigorous microeconomic foundations, the absence of which was the fatal flaw of the Keynesian consensus of the 1960s.” His final assessment is that real business cycle theory “has served the important function of stimulating and provoking the scientific debate” but “will, I predict, ultimately be discarded as an explanation of observed fluctuations.”