Chris Sims’ contributions to macroeconometric methodology are well-known and justly celebrated, including, of course, by the Nobel Prize that he shared with Tom Sargent in 2011. But while Chris is most often associated with advocacy of atheoretical data characterisation techniques, such as the vector autoregressions that he illustrated in Sims (1980), he maintained throughout his career an active interest in improving our understanding of the structural mechanisms underlying observed macroeconomic dynamics. And Chris’ contributions to structural macroeconomic modelling have been crucial for the development of the field, even if less widely discussed.

Indeed, Chris’s contributions to structural modelling in the 1980s and 1990s were decisive for the development of what have come to be known as New Keynesian DSGE models. This has been true not only because of his constant insistence on the importance of simultaneously modelling the co-evolution of a large number of economic time series, rather than expecting that correct causal inferences could be drawn from single-equation regressions. Chris’s dogged pursuit of a correct structural interpretation of the relationship between monetary variables and aggregate economic activity directly influenced the way this issue was treated in the NK DSGE models; this in turn was one of the most important ways in which those models changed the field’s view of macroeconomic fluctuations.

The role of monetary policy as an explanation of fluctuations in business activity had been a topic of intense discussion since the 1960s, most directly as a result of the studies of Milton Friedman and Anna J. Schwartz, documenting the correlation between fluctuations in monetary aggregates and measures of real activity over many decades. Friedman and Schwartz argued that these co-movements should be interpreted as evidence of a causal effect of changes in the money supply on real activity, and on that basis concluded that erratic monetary policy had been the main cause of observed business cycles. But others, such as Tobin (1970), argued that co-movement of money and aggregate activity could equally well be interpreted as resulting from reverse causation, with variations in real activity due entirely to other factors and variations in the money supply then following them, owing to central-bank accommodation of the variation in money demand resulting from variable economic activity.

This posed a challenge as to whether empirical evidence could distinguish between these two very different interpretations of money-income correlations, and the effort to meet that challenge through more sophisticated econometrics was one of the continuing preoccupations of Chris Sims’ career. In his first important contribution on this topic (Sims 1972), he examined the timing relationship between money and output using the novel tool of Granger causality, showing that there was a Granger-causal relationship in the Granger from money to output, and not the reverse. This was widely interpreted as support for the Friedman-Schwartz view, and hence for a monetarist view of the ultimate source of business fluctuations.

But while the result was celebrated – seeming as it did to confirm the conventional wisdom of the time – Chris himself was not content to leave the matter there. In his subsequent work he continued to return to the problem, insisting on the subtlety of the identification problem involved given the plausibility of an endogenous component to the variation in monetary aggregates at business-cycle frequencies, and on the importance of looking at the co-movements of multiple variables in order to draw more credible conclusions.

An important turning point was a short paper that he published in 1986 in the Quarterly Review of the Minneapolis Fed (Sims 1986). The paper was a response to criticisms of the usefulness of VAR models in policy evaluation. While much of the paper is devoted to general issues of economic and econometric methodology, as an application it revisits the question of the importance of monetary policy disturbances for real activity, and the channels through which they have their effects. In one of the earliest applications of what came to be known as ‘structural VAR’ methodology, Chris proposes a new way of identifying the effects of a monetary-policy shock, understood to mean an unforecastable innovation in the intercept of a central-bank reaction function.

There were clear limits to the extent to which such an exercise could provide answers to questions about how inflation and output dynamics would be different under a systematic policy rule of a different kind than had historically been followed, as emphasized by the ‘Lucas critique’ (Lucas 1976). But it could nonetheless contribute importantly to the assessment of alternative causal interpretations of the historical experience. In particular, Chris emphasised that the results from a plausible set of identifying assumptions implied “that the more naïve versions of both rational expectations monetarism and conventional monetarism are mistaken” (p. 15). Such conclusions about the mechanisms at play in the period for which the VAR was estimated could then reasonably be used in choosing a structural model with which to analyse the effects of counterfactual policies.

The structural VAR model proposed in Sims (1986), and further developments of it in Chris’s work in the 1990s (e.g. Sims 1992, Leeper and Sims 1994, Leeper et al. 1996) contained several elements that were crucial for the development of New Keynesian DSGE models later in that decade. Monetary policy shocks were no longer identified by the unforecastable component of some monetary aggregate; instead, innovations in the quantity of money were interpreted as being determined by shocks to both money supply and demand, in a simultaneous-equations system. Innovations in a short-term nominal interest rate were also an important source of information about monetary policy, given a central-bank reaction function that would raise the interest rate in response to increases in money demand (though not so sharply as to completely eliminate any increase in the quantity of money). The identified monetary policy shocks were then the unforecastable shifts in the intercept of this reaction function, rather than innovations in either the interest rate or money alone. 

In Sims (1986), the interest-rate indicator was a short-term Treasury bill rate. Following the work of Bernanke and Blinder (1992), the New Keynesian literature tended to emphasise the federal funds rate as the interest rate most directly controlled by the Fed, and hence the variable in terms of which it was most useful to specify a central-bank reaction function. And in the VAR dynamics to which Rotemberg and Woodford (1998) fit their estimated New Keynesian DSGE model, the money supply is not even included as an indicator variable for that reaction function. Nonetheless, their approach followed Sims in identifying monetary policy shocks not with funds rate innovations as such, but with the part of those innovations that could not be accounted for by innovations in the other variables affecting the central-bank reaction function within the quarter. 

An important conclusion from Chris’ VAR studies was that when monetary policy shocks were identified in this way, they were found to account for only a small fraction of the variability in aggregate economic activity. But this did not mean that monetary policy doesn’t matter for economic stability – only that it matters mainly through the implications of systematic monetary policy for the transmission of the effects of real disturbances. And to understand those effects of systematic policy, measurement of the effects of identified monetary policy shocks can be valuable, though not because the shocks themselves are important. Instead, the nominal rigidities that are needed to explain the effects of those shocks are important to measure, given their implications for the propagation of other kinds of disturbances. For that reason, the choice of model parameters to fit the estimated impulse responses to the identified monetary policy shock in a structural VAR became the key to the estimation strategies of both Rotemberg and Woodford (1998) and Christiano et al. (2005).

In other respects, Chris’ work not only foreshadowed later developments in the New Keynesian literature, but pointed to themes that that literature has yet to fully develop, though arguably it should. Standard New Keynesian DSGE models like those just mentioned (or the more fully developed econometric model of Smets and Wouters 2007) are still ‘monetarist’ models in that they model government policy as influencing aggregate demand solely through monetary policy, and they imply that inflation should in principle be highly controllable through appropriate adjustment of the central bank’s monetary instruments. 

Instead, Chris’ work on the “fiscal theory of the price level” (Sims1994) assigned an equal role to systematic fiscal policy in the determination of both real and nominal aggregate variables, and stressed the constraints that fiscal policy places on what it is possible for monetary policy to achieve. This was an increasingly frequent theme in his late work (e.g. Sims 2024). Given recent fiscal trends in the US and many other countries, these constraints seem likely to be even more central to the policy debates of the next decades. It is a great loss to our field that we no longer have Chris to lead the way in facing the challenges for macroeconomic modelling that these developments pose.

Chris Sims’ contributions to my own development as an economist have been immense. The seriousness of his commitment to getting answers right, even if this required decades of repeated reformulations, and his aversion to easy but superficial solutions were a source of inspiration, and a constant goad to try to do better. I treasure the many conversations we had over the years in which I have tried to follow in his footsteps, wishing only that it were possible to pursue them further.

References

Bernanke, B S and A S Blinder (1992), “The Federal Funds Rate and the Channels of Monetary Transmission,” American Economic Review 82: 901-921. 

Christiano, L J, M Eichenbaum, and C L Evans (2005), “Nominal Rigidities and the Dynamic Effects of a Shock to Monetary Policy”, Journal of Political Economy 113: 1-45.

Leeper, E M and C A Sims (1994), “Toward a Modern Macroeconomic Model Usable for Policy Analysis,” in S Fischer and J J Rotemberg (eds), NBER Macroeconomics Annual 1994, MIT Press.  

Leeper, E M, C A Sims, and T Zha (1996), “What Does Monetary Policy Do?” Brookings Papers on Economic Activity 1996-2: 1-78.  

Lucas, Jr, R E (1976), “Econometric Policy Evaluation: A Critique,” in K Brunner and A Meltzer (eds), Carnegie-Rochester Conference Series on Public Policy, Vol. 1, Elsevier Press.

Rotemberg, J J and M Woodford (1976), “An Optimization-Based Econometric Framework for the Evaluation of Monetary Policy,” in B S Bernanke and J J Rotemberg (eds), NBER Macroeconomics Annual 1997, MIT Press.

Sims, C A (1972), “Money, Income, and Causality,” American Economic Review 62: 540-552.  

Sims, C A (1980), “Macroeconomics and Reality,” Econometrica 48: 1-48. 

Sims, C A (1986), “Are Forecasting Models Usable for Policy Analysis?” Federal Reserve Bank of Minneapolis Quarterly Review 10(1): 2-16.  

Sims, C A (1992), “Interpreting the Macroeconomic Time Series Facts: The Effects of Monetary Policy,” European Economic Review 36: 975-1000.  

Sims, C A (1994), “A Simple Model for Study of the Determination of the Price Level and the Interaction of Monetary and Fiscal Policy,” Economic Theory 4: 381-399.

Sims, C A (2024), “Origins of U.S. Inflation,” AEA Papers and Proceedings 114: 90-94.

Smets, F and R Wouters (2007), “Shocks and Frictions in US Business Cycles: A Bayesian DSGE Approach,” American Economic Review 97: 586-606.

Tobin, J (1970), “Money and Income: Post Hoc Ergo Propter Hoc?”, Quarterly Journal of Economics 84: 301-317.