The Nature and Evolution of Macroeconomics

The subject matter of macroeconomics is usually divided into two distinct areas: (1) the analysis of long-run economic growth, which is the main focus of Smith [1776] and other classical economists, and (2) the analysis of aggregate fluctuations, which is the main focus of Hume [1752], subsequent monetary and other business cycle theorists, and, of course, Keynes [1936].

1.1.1 Pre-Keynesian Macroeconomics

Macroeconomics did not exist as a separate field of economics before the publication of the General Theory in 1936. However, at least three important theoretical streams foreshadowed it.

One was the analysis of long-run economic growth, an important concern of classical economists, such as Smith, Malthus, Ricardo, and Mill. The classical economists sought to explain economic growth in terms of population growth; the accumulation of capital; and the increase in the efficiency of production, as determined by the division of labor and technical progress. These factors interacted with the availability of land, a factor of production that was assumed to be in fixed supply. The classical theory of economic growth, focusing on the production and distribution of income, is essentially macroeconomic in nature.²

The second stream of pre-Keynesian macroeconomics is monetary theory. Monetary theory was quite advanced even before the emergence of classical economics. Hume [1752] provides important examples of advanced monetary analysis more than 20 years before Smith. By the early twentieth century, monetary theory had established the quantity theory of money, the classical dichotomy between real and nominal variables, and some monetary explanations of the business cycle. The quantity theory of money suggests that increases in the money supply lead, at least eventually, to equiproportional increases in the general level of prices.

Monetary theories of the business cycle attribute the business cycle to the short-term real effects of monetary factors interacting with the gradual adjustment of wages and prices (Hume [1752], Thornton [1802]), with misperceptions between changes in relative prices and the price level (Mill, [1833, 1848]) and with temporary deviations of the interest rate from its equilibrium value (Fisher [1896], Wicksell [1898], von Mises [1912]). Various monetary schools of thought had evolved by the latter part of the nineteenth century and still existed in the early part of the twentieth century.³

The third stream foreshadowing macroeconomics consists of various types of real business cycle theories. Apart from the dominant monetary theories, there were alternative theories that attempt to explain aggregate fluctuations in terms of overinvestment, underconsumption, excess indebtedness, psychology, technology, or harvest and agricultural cycles. These theories are also essentially macroeconomic in nature.⁴

The term “macroeconomics” predates the General Theory and has to be credited to Frisch [1933, p. 2], who defines it as follows:

When we approach the study of business cycles with the intention of carrying through an analysis that is truly dynamic and determinate, we are naturally led to distinguish between two types of analyses: the micro-dynamic and the macro-dynamic types. …The macro-dynamic analysis … tries to give an account of the fluctuations of the whole economic system taken in its entirety.

In any event, the General Theory, as codified in the IS-LM framework of Hicks [1937], and later by Modigliani [1944], Samuelson [1948], Hansen [1949], Patinkin [1956], and others, eventually prevailed over previous approaches during the 1940s and the 1950s and was pivotal in the establishment of macroeconomics as a separate field of economics.⁵

Macroeconomics gradually integrated the three theoretical streams mentioned above and experienced explosive growth during the rest of the twentieth century.

1.1.2 Classical and Keynesian Macroeconomics

Following the Keynesian revolution of the 1930s, macroeconomics originally evolved with little reliance on underlying microeconomic theory. For the most part, in the aftermath of the trauma of the Great Depression and the scathing attack by Keynes [1936] on classical economics, macroeconomics based on solid microeconomic principles was dismissed as classical macroeconomics. In the increasingly dominant paradigm of Keynesian macroeconomics, during the 1950s and the 1960s, most key aggregate relations (such as the consumption function, the investment function, and the relation between inflation and unemployment) were postulated rather than derived from explicit choice–theoretic microeconomic foundations. Surprisingly, this applied not only to the short-run Keynesian model of aggregate fluctuations but also to models of long-run economic growth.

To a large extent, the instability and knife-edge conditions characterizing the early post-Keynesian models of economic growth (those by Harrod [1939] and Domar [1946]) were due to their unsatisfactory microeconomic foundations, such as the postulates of a constant savings rate and the assumed absence of long-run substitution possibilities between capital and labor in the production of goods and services.

The neoclassical Solow [1956] and Swan [1956] model of economic growth—which was based on a much more general production function allowing for substitution between capital and labor—also relied on a postulated Keynesian consumption function that was based on the assumption of the General Theory that consumption is an exogenous fraction of current income.

It took some time before Cass [1965] and Koopmans [1965] rediscovered and extended the Ramsey [1928] representative household model of optimal savings and reestablished the link between growth theory and optimizing households.

Both the representative household model and the overlapping generations model are dynamic general equilibrium models with explicit microeconomic foundations and are widely used in macroeconomics to this day.⁷

The Keynesian approach to aggregate fluctuations, which became totally dominant in the 1950s and the 1960s, suffers from insufficient microeconomic foundations to an even greater extent than growth theory. This applies to both theoretical models (such as the IS-LM framework of Hicks [1937], as adapted by Hansen [1949] and the models of Samuelson [1939] and Modigliani [1944]) and econometric models (such as the Klein [1950] and Klein and Goldberger [1955] models).⁸

For example, the IS-LM framework of Hicks [1937] is essentially a static short-run general equilibrium model of income and interest rate determination, based on ad hoc General Theory postulates, such as the positive consumption-income relation, the negative investment–interest rate relation, and liquidity preference.

Even the multiplier-accelerator model of Samuelson [1939], probably the most influential early dynamic business cycle model based on Keynesian principles, relies on a simple postulated consumption function, with consumption being a linear function of past income and investment a constant multiple of the change in consumption. The marginal propensity to consume out of past income defines the multiplier, and the marginal propensity to invest, following a change in consumption, defines the accelerator. Yet neither the multiplier nor the accelerator was derived from an optimizing microeconomic model for households and firms.⁹

The same also applied, although to a lesser extent, to the so-called neoclassical synthesis, which is a combination of the IS-LM framework with an aggregate short-run supply function that depends on the assumption of short-run rigidity of nominal wages and prices.¹⁰

This state of affairs was of significant concern to many economists, including the protagonists of the development of the Keynesian models themselves, who were unhappy with the weakness of the microeconomic foundations of many of the postulated macroeconomic relations.

1.1.3 Microeconomic Foundations of Macroeconomics

Beginning in the 1950s, the various attempts to provide better microeconomic foundations for macroeconomics gradually started to bear fruit.

Modigliani and Brumberg [1954] and Friedman [1957] provide dynamic microeconomic foundations for the consumption function, based on intertemporal considerations. From these efforts came the life cycle and permanent income theories of consumption, which differ from the simple static Keynesian consumption function.¹¹

Jorgenson [1963] introduces the flexible accelerator model of investment, based on profit maximization by firms and the assumption of adjustment costs for the capital stock. His contribution and that of Tobin [1969] led to the modern optimizing q theories of investment.¹²

Baumol [1952], Tobin [1956], Friedman [1956], Patinkin [1956], Samuelson [1958], and others derive the demand for money from the optimizing behavior of households and firms.¹³

More importantly, Patinkin [1956] sought to base the whole model of the neoclassical synthesis on a better microeconomic footing, consistent with developments in Walrasian general equilibrium theory. The program of Patinkin was carried forward by Clower [1965], Leijonhufvud [1968], and the so-called non-Walrasian equilibrium modeling of Barro and Grossman [1971], Muellbauer and Portes [1978], and others.

These are only some of the early attempts to provide microeconomic foundations for the postulated key relations of Keynesian macroeconomic models.

The most celebrated case of an empirical relationship with inadequate microeconomic foundations is probably the Phillips curve. This was an inverse empirical relation between inflation and unemployment, discovered by Phillips [1958]. Initially, very little theory underpinned the Phillips curve, which was interpreted simply as an adjustment equation for wages, depending on the excess demand for labor.

However, in the late 1960s, the Phillips curve appeared to break down. Rises in inflation resulted in only temporary reductions in unemployment. Phelps [1967] and Friedman [1968] were able to explain the breakdown of the Phillips curve, and even to anticipate it to a certain extent, in terms of shifts in inflationary expectations. They used the first rudimentary optimizing models of inflation and unemployment, which resulted in the so-called expectations-augmented Phillips curve and the natural rate hypothesis.¹⁴

An important research effort, seeking to provide firm dynamic microeconomic foundations for the Phillips curve, followed almost immediately afterward. The contributions in Phelps [1970] kickstarted this program. The rational expectations revolution followed suit, when Lucas [1972] applied the rational expectations hypothesis of Muth [1961] to a general equilibrium model of the Phillips curve. Up until then, the standard expectations hypothesis used in macroeconomics was the hypothesis of adaptive expectations.¹⁵

Gradually, the focus shifted to dynamic stochastic general equilibrium (DSGE) models of aggregate fluctuations, based on dynamic optimization by households and firms, following the ideas of Lucas [1977] and the important early such model by Kydland and Prescott [1980, 1982]. These models were initially in the classical tradition and led to what is now termed new classical macroeconomics, or real business cycle theory.¹⁶

Alternative DSGE models were also developed in the monetary and Keynesian tradition of gradual adjustment of prices and nominal wages. These alternative models emphasized both real and nominal distortions, such as labor market imperfections and nominal wage contracts, or imperfect competition and staggered pricing in product markets. Mankiw and Romer [1991] is an early collection of papers in what is now termed new Keynesian macroeconomics. These models can account for fluctuations caused by monetary factors as well as real factors. They can also justify a stabilizing role for monetary policy not only for inflation but also for fluctuations in real output and employment.¹⁷

1.1.4 Deterministic and Stochastic Dynamic General Equilibrium Models

Both the new classical and the new Keynesian approaches coexist today, in the context of what Goodfriend and King [1997] have termed the “new neoclassical synthesis.” This synthesis is based on DSGE models, which may or may not satisfy the conditions for optimality of the Arrow-Debreu model. They share many common elements but are also characterized by important differences with regard to their accounts of aggregate fluctuations and their implications for monetary policy.¹⁸

In a different development, after a long pause, growth theory has displayed significant progress since the 1980s in the context of what has been called “new growth theory.” This work has focused on the exploration of the implications of increasing returns to scale, the role of human capital accumulation, and endogenous technical progress. A more recent literature has emphasized the interactions between fertility and technical progress as a choice between the quantity and quality of children. This has resulted in the development of the so-called unified growth theory, which, unlike previous theories, can also explain the transition from stagnation to growth in the early nineteenth century. Yet another stream in the literature emphasizes the role of institutions in providing appropriate incentives for the accumulation of physical and human capital and for research and development that leads to technical progress.¹⁹

In seeking to explain long-run economic growth and fluctuations in aggregate economic activity, modern macroeconomics is dynamic. The element of time is indispensable for understanding and explaining both types of phenomena. In seeking to explain aggregate fluctuations, modern macroeconomics is also stochastic: It explains aggregate fluctuations in terms of the response of dynamic economic systems to random disturbances. Stochastic growth models have also been developed. Hence, both the new neoclassical synthesis and the new growth theory combine dynamic and stochastic elements.²⁰

The dynamic stochastic approach to aggregate fluctuations follows a tradition that was also founded in the 1930s, by mathematical economists and statisticians, such as Frisch [1933] and Slutsky [1937]. This tradition, which initially evolved independently of the General Theory, was subsequently followed and extended in various directions by econometricians, such as Tinbergen [1937], Haavelmo [1944] and the Cowles Commission, and Burns and Mitchell [1946]. The development of early Keynesian macroeconometric models by Klein [1950], Klein and Goldberger [1955], and others applied this tradition to the framework of the General Theory.²¹

The way in which modern macroeconomics approaches and analyzes aggregate fluctuations owes a lot to the following important observation by Lucas [1977, pp. 9–10]:

Technically, movements about trend in gross national product in any country can be well described by a stochastically disturbed difference equation of very low order. These movements do not exhibit uniformity of either period or amplitude, which is to say, they do not resemble the deterministic wave motions which sometimes arise in the natural sciences. Those regularities which are observed are in the co-movements among different aggregative time series. …One is led by the facts to conclude that, with respect to the qualitative behavior of co-movements among series, business cycles are all alike. To theoretically inclined economists, this conclusion should be attractive and challenging, for it suggests the possibility of a unified explanation of business cycles, grounded in the general laws governing market economies, rather than in political or institutional characteristics specific to particular countries or periods.

This observation by Lucas, based on the foundations laid out by Frisch [1933], Slutsky [1937], and Burns and Mitchell [1946], brought about significant changes in the way in which all schools of thought in modern macroeconomics approach and try to explain aggregate fluctuations.

The traditional Keynesian macroeconomic and macroeconometric models, from the 1950s to the 1970s, were deemed to have weaknesses in the detailed study of aggregate fluctuations and the impact of monetary and fiscal policy in relation to the criteria of Lucas. The most important of these weaknesses was that the macroeconomic relationships assumed in traditional models were not explicitly drawn from well-defined microeconomic foundations, based on intertemporal optimization on the part of households and firms. Therefore, one could not easily interpret their parameters and be confident in their stability. This became the basis of the Lucas critique of econometric policy evaluation. Lucas [1976, p. 41] concluded that

Given that the structure of all econometric model consists of optimal decision rules of economic agents, and that optimal decision rules vary systematically with changes in the structure of series relevant to the decision maker, it follows that any change in policy will systematically alter the structure of econometric models.

Lucas arrived at this conclusion after having demonstrated the fragility of traditional econometric models of aggregate consumption, investment, and the Phillips curve. Pursuing this critique—and the quest for dynamic microeconomic foundations in macroeconomic models—the macroeconomics of aggregate fluctuations shifted to the study of DSGE models with explicit dynamic microeconomic foundations.

Modern macroeconomics is now almost entirely based on such dynamic general equilibrium models, which may be either deterministic or stochastic. For the study of economic growth, the main models are variants of the representative household model (due to Ramsey [1928], Cass [1965], and Koopmans [1965]) and overlapping generations models (such as the Diamond [1965] and Blanchard [1985]–Weil [1989] models). These are chiefly deterministic dynamic general equilibrium models, although stochastic elements can be added to them. Models used for the study of aggregate fluctuations combine elements from both new classical and new Keynesian DSGE models to form the basis of the new neoclassical synthesis. Naturally, such models form the backbone of the present book. I present the main theories of economic growth and aggregate fluctuations through a sequence of such dynamic models, based on intertemporal optimization on the part of economic agents. For the most part, these models are transformed into linear or log-linear systems of equations. In chapter 22, we also discuss an alternative approach to aggregate fluctuations based on nonlinear overlapping generations models that result in endogenous cycles, self-fulfilling prophecies, and sunspots (Azariadis [1981], Azariadis and Guesnerie [1986], Benhabib and Farmer [1999]). For completeness, I also discuss some of the important precursors to these dynamic general equilibrium models.²²

The models presented and analyzed in this book are treated as tools for understanding the main macroeconomic phenomena of long-run economic growth, aggregate fluctuations, inflation and unemployment, and the role of monetary and fiscal policies. The book highlights both their potential strengths as well as their limitations.

It is worth keeping in mind that modern macroeconomics is not based on a single, generally accepted, all-encompassing model. For this reason, this book is eclectic and treats macroeconomics as applied and policy-oriented general equilibrium analysis, based on various alternative, relatively simple aggregate dynamic models. We examine a plurality of models, each of which is suitable for investigating specific issues and addressing specific questions but may be unsuitable for other issues or questions.²³

Some key unifying principles are found in the models that we adopt. The most important of these principles is the assumption that economic agents base their decisions on intertemporal optimization of some well-defined objective function under appropriate constraints. Thus, for the most part, we rely on dynamic general equilibrium models with explicit intertemporal microeconomic foundations. Where there are theoretical disagreements, alternative approaches are juxtaposed, their pros and cons are analyzed, and their compatibility with the empirical evidence is also briefly discussed.

Before we turn to the theory and the models themselves, it is worth looking at the key empirical facts concerning long-run economic growth and aggregate fluctuations. These key facts are what macroeconomics seeks to explain and account for.

We start with some of the key facts about long-run economic growth and then move on to some of the key facts about aggregate fluctuations. Additional facts are also presented as we move to particular models in the relevant chapters and the specific issues these models seek to explain. Knowledge of these key facts will facilitate the process of evaluating the relevance and usefulness of the theoretical models in the rest of this book.²⁴

1.2