Detecting Fiscal-Monetary Causes of Inflation

Introduction

Latin American countries have experimented with a wide variety of macroeconomic policies and experienced diverse and sometimes adverse consequent outcomes. The Monetary and Fiscal History of Latin America (MFH) project has collected and organized systematic evidence consisting of comparable data sets from these varied historical experiences to construct a knowledge base for studying the origins and effects of alternative monetary and fiscal policies. Instances of these diverse policies were themselves caused by various political and economic events, and so they can’t easily be viewed as “experiments” purposefully conducted to learn the general equilibrium effects of alternative policies. Indeed, the study of macroeconomic policies explores not only their direct effects on individuals’ behaviors and their indirect effects to altered market prices but also the responses of future policies to changes in current policies. Such interactions are central focuses of policy analysis.

While many economists emphasize the importance of evidence-based policy analysis, the evidence won’t speak for itself; we require a conceptual framework to organize and interpret evidence. Decades ago, Koopmans (1947), while at the Cowles Commission at Chicago, wrote “Measurement without Theory,” a critical review of extensive research by Burns and Mitchell (1946) that measured business cycles from macroeconomic data. Koopman’s review sparked a debate (for instance, see Vining 1949) that continues to this day as we struggle to put data science and machine learning methods to scientific use. Marschak (1953), Hurwicz (1966), and Lucas (1976) are classic expositions of the case for using structural econometric models to do policy analysis.

The architects of the MFH project, Tim Kehoe and Juan Pablo Nicolini, are keenly aware of the interplay between evidence and theory. They aimed to design the project to facilitate the use of multicountry data to compare and contrast alternative models of the fiscal and monetary interactions.

As macroeconomists, we find this study to be particularly promising because of the varied institutional structures, policy experiences, and macroeconomic outcomes presented. The countries have different degrees of separation between responsibilities for designing and executing monetary and fiscal policies. These separation conventions influence how monetary and fiscal policies are ultimately coordinated and the extent to which one policy authority dominates the other. Such institutional conventions matter for alternative models of macroeconomic impacts. Whether fiscal or monetary policies are dominant or the extent to which they act in unison with common ambitions can have important consequences for macroeconomic performance. Studies like the MFH project promote understanding of the impacts of monetary-fiscal policy interactions.

Intellectual History

As background, we recall how Milton Friedman encouraged his students Eugene Lerner (Lerner 1956) and Phillip Cagan (Cagan 1956) to study big inflations that accompanied the Civil War in the United States and that followed both World Wars I and II.1 Friedman recognized that these episodes contained sources of data variations that would let Lerner and Cagan isolate fundamental monetary-fiscal causes of inflation and private behavioral responses to it that can be confounded by many other forces that also affect the price level. Thus, Friedman and his students aimed to sidestep the complex interactions of monetary and fiscal policy and avoid taking account of various general equilibrium effects by focusing on data for which monetary policy would dominate macroeconomic effects.

Specifically, Friedman’s students sought:

1. 1. a single behavioral relationship—an aggregate demand function for money linking real balances inversely to an expected rate of inflation;
2. 2. measures of the supply of nominal money balances;
3. 3. links between time series of measured inflation and an estimate of the public’s expected rate of inflation suitable for plugging into the demand function for real balances.

Friedman’s hunch was that the episodes studied by Lerner and Cagan could be treated as though they were “natural experiments.” Specifically, explosive money supply series were “uncaused events” whose sources need not be analyzed in order to understand the main force driving the price level and real balances. During the episodes studied by Lerner and Cagan, the three items on Friedman’s list would overwhelm a long list of confounding forces that in tranquil times also affect the price level and make the quantity-theoretic sources of inflation more difficult to detect.

The MFH project led by Kehoe and Nicolini aims to extend the quantity- theoretic tradition pursued by Friedman and his students in ways that can help us to understand the histories of inflation in a set of Latin American countries. All of the countries studied experienced a variety of inflationary episodes, though usually at rates substantially below those in the short European hyperinflation episodes studied by Cagan.2 And while the three identification pillars of Friedman’s students’ analyses play roles in these new studies, variations in the rates of money creation and inflation are sufficiently lower that other forces confound their influences. So studies in the MFH project also pay attention to other data that the researchers hope will allow them to detect those other causes of inflation.

Explicitly dynamic equilibrium analyses of economies are vital for understanding how macroeconomic policies operate. The analyses avoid the “measurement without theory” criticism stated by Koopmans. While this approach makes macroeconomic policy analysis challenging, results can be enlightening. We should not expect the data and analyses from this project to settle all modeling and measurement challenges, although the authors and architects of the study have taken important steps forward.

The main hypothesis explored by Kehoe and Nicolini is that sustained inflation rates, as opposed to short bursts of hyperinflation, are outcomes of particular monetary and fiscal regimes and the associated (perhaps rational) expectations about future policy actions that they cause. This is what led Kehoe and Nicolini to stress a common measurement of the consolidated budget constraint faced by monetary and fiscal authorities, together with the institutional backgrounds in which they are managed. Importantly, interrelated monetary and fiscal policies sometimes lead to indeterminate outcomes in the form of multiple equilibria that offer potential sources of variations in macroeconomic outcomes. One example is that economies can be pulled in different directions at different points in time. For example, Sargent (1999) and Sargent, Williams, and Zha (2009) illustrated how a multiplicity of self- confirming equilibria in conjunction with an adaptive learning mechanism present a dynamic pull toward and escape from these equilibria. Finally, a point that is usually deemphasized or ignored in models in the rational expectations tradition is that people inside economic models (e.g., consumers, entrepreneurs, and even policymakers) are exposed to macroeconomic ambiguity as they wrestle with uncertain policies. (For example, see the discussion in Hansen 2014.) Sometimes adaptive learning is motivated by agents’ awareness of potential model misspecifications. It is worth noting that there are also “control theory” counterparts that can be used to understand how people inside our models perceive the complex uncertainty that they confront and that affect both markets and policymaking. We will elaborate on some existing work that features adaptive learning as a way to model potentially shifting environments.

This introduction presents and explores a list of theories about forces that help us to understand equilibrium responses to alternative macroeconomic policy configurations. To interpret evidence, we find it compelling to push beyond the identification approach used in the empirical monetarist approach of Friedman and his former students Cagan and Lerner. This approach is needed to refine inferences about those basic forces to be drawn from the types of episodes studied by the team of researchers contributing to the MFH project. In so doing, model builders and econometricians are compelled to:

1. 1. refine models of expectations;
2. 2. make components endogenous that Friedman and his colleagues took as exogenous;
3. 3. add more “realistic” components of government budget constraints, such as risk-free real debt, risk-free nominal debt, and defaultable debt.

Challenges to “Exogenous” Money Supply

In this section, we review three empirical-econometric paths that refined and extended the money demand model used in the classic studies by Friedman and his former students.

Inflation Theories under Cagan Demand

We now explore some initial extensions of the Friedman-Cagan money demand approach with explicit fiscal monetary interactions. To begin, suppose that the government budget constraint is financed entirely by printing money. This happened to be one of Friedman’s recommendations for coordinating monetary and fiscal policy (see Friedman 1948). Interestingly, Friedman wrote that proposal a few years before the accord between the Treasury and Federal Reserve that granted the Federal Reserve independence from the Treasury. Friedman’s proposal recommends complete fiscal dominance over monetary policy. Within such a monetary-fiscal setup, we explore alternative ways of modeling private-sector beliefs.

Models without Full Debt Monetization

In the preceding models, government deficits are financed entirely by money creation. This structure serves as a pedagogically revealing example of the interplay between monetary and fiscal policy, but it is too special for understanding macroeconomic policies and their outcomes in many countries at many times. The unpleasant monetarist arithmetic of Sargent and Wallace (1981) adds one-period government debt to a model with perfect foresight and a demand for real balances like that of Cagan’s. This simple setting frames the need to coordinate a monetary policy that determines a rate of money creation process with a fiscal policy that determines a net-of-interest government deficit process. Because seigniorage is a source of real government revenues and hence affects the equilibrium present value of these revenues, the “independence” of monetary authorities is a fiction (or said more politely, a “convention”). Sargent and Wallace stated conditions on fiscal policy and equilibrium real rates of interest under which a monetary authority could fight inflation in the short run only by making it worse in the long run.

The Sargent and Wallace unpleasant arithmetic also opened the door to a “fiscal theory of the price level” developed subsequently by Sims (1994), Woodford (1995), Leeper (1991), and others (See Leeper and Leith 2016 for a comprehensive survey and Loyo 1999 for application to Brazilian inflationary episodes.) In a simple instance, imagine an environment in which real discount factors (potentially stochastic) are determined by an external economy and in which a real net-of-interest government surplus series is specified exogenously. The real value of debt is pinned down by a present-value relation. Since nominal debt is predetermined, the nominal price follows from a formula for the equilibrium present-value relation. This gives a rather stark example of the impact of fiscal policy on price determination. Of course, this is a (typically too) simple specification of the dynamic economic system. More generally, we expect more endogeneity, and this present-value relation is one among a system of equations that must hold in equilibrium.4

As in the example economies with full monetization, there is a multiplicity of equilibria, and the details of the monetary and fiscal rules matter and can appear as alternative types of regimes in which either monetary policy or fiscal policy dominates. By allowing for more than one-period debt, Cochrane (2001) explored policy ramifications for the full term structure. Recently, Chen, Leeper, and Leith (2015) and Bianchi and Ilut (2017) contributed an econometric exploration of regime-shift models between so-called fiscal dominance and monetary dominance. As an alternative, the Marcet and Sargent (1989a) type approach could be employed in which the private sector embraces adaptive learning, opening the door to expectations-based escape dynamics with different forces pushing toward alternative equilibria.

Overall, these models point out the potential importance of the rules of engagement between fiscal and monetary authorities, as well as the crucial role of expectations formation in determining macroeconomic outcomes. They further demonstrate the need to study monetary and fiscal history together, as well as the value of a common measurement framework for the interpretation of the varied economic experiences in Latin America.5

Models with Risky Nominal Government Debt

Pushing fiscal policy to the forefront puts a focus on the tools and consequences of debt management. To frame this discussion, we start with the classic benchmark model of Lucas and Stokey (1983) that describes an optimal policy analysis under commitment when there exist a rich (complete) set of financial markets. Lucas and Stokey explored monetary-fiscal interactions in several revealing ways. Parts of the paper are tied to the Ramsey enterprise of constructing an optimal monetary-fiscal policy mix under well-posed rules of the game. The model has become a standard reference point for analyzing the consequences of risky government debt and for demonstrating how inflation can make nominal government debt mimic risky real government debt when managed appropriately. But because they recover a version of a Friedman rule of a zero nominal interest rate as a Ramsey policy, Lucas and Stokey’s normative analysis requires modification in developing a positive description needed to understand various Latin American high-inflation episodes.

Extensions of potential value for interpreting evidence include:

1. 1. introducing explicit forms of market incompleteness;
2. 2. allowing for default;
3. 3. adding tractable forms of heterogeneity across sectors, external investors, and individuals within the counties under investigation.

Such extensions put inflation and deflation into the arsenal that a Ramsey planner could use to redistribute among nominal debtors and creditors in response to macro shocks. Moreover, the planner could denominate debt in either local or foreign currency as part of the policy design.6

Phillips Curve Dynamics

The classic Friedman-Lerner-Cagan studies shut down Phillips curve dynamics by taking aggregate supplies of goods and aggregate employment as exogenous. A number of the subsequent contributions mentioned previously followed in this tradition by shutting down exploitable or permanent trade-offs between inflation and real outputs (e.g., the unpleasant monetarist arithmetic of Sargent and Wallace [1981] and the Lucas and Stokey [1983] investigation of state-contingent debt and taxation). Macroeconomic analyses of developed economies often embrace “sticky prices” to activate exploitable Phillips curve dynamics that at low or moderate inflation rates disguise the inflationary forces isolated by Cagan and Lerner.

Most current research modeling of Phillips curve dynamics in developed economies relies on Calvo’s (exogenously specified) price-setting mechanism. Imposing this structure dramatically simplifies calculations, often with little sacrifice of insights that would come from a deeper theory of price stickiness. Nevertheless, in the case of persistent high inflation or large nominal shocks, Calvo’s modeling device can have quantitatively counterfactual implications. Moreover, the Calvo price-setting Poisson coefficients cannot plausibly be transported across economies with very different monetary-fiscal policies. Embracing a more primitive starting point to model price setting opens the door to a better understanding of economies that experience higher levels of inflation. An attractive alternative is to use a model price setting with menu costs, as in Sheshinski and Weiss (1977), or its general equilibrium version with idiosyncratic shocks, as in Golosov and Lucas (2007). Alvarez et al. (2019) obtained theoretical predictions for this class of models and showed that they are in line with data for large inflation rates, including findings from earlier research by Gagnon (2009) based on evidence from Mexico. Exploring the conceptual underpinnings of price sluggishness promises to provide a unified framework for better understanding the evidence and its implications across the varied inflationary experiences in Latin America as studied in the MFH.

Putting Things Together

Cagan (1956) was able to do a good job of explaining a set of spectacular hyperinflations by using a simple model. Relative to Cagan’s model, we have added complications that seem vital for understanding Latin American macroeconomic histories. Because the monetary-fiscal events studied by the Kehoe-Nicolini team are less extreme than those studied by Cagan, the Kehoe-Nicolini team cannot neglect the confounding forces that Cagan could ignore. Prime among these are government debts. For example, by the time the German hyperinflation got rolling, the German government had defaulted on virtually all of its domestic debt, so for better or worse, Cagan could ignore government debt.

For many of the episodes studied by the Kehoe-Nicolini team, government debts and how and whether they were paid, rescheduled, or defaulted on are big parts of the story. But limitations on how governments and central banks account for government debts present substantial measurement difficulties. The budget constraints of macroeconomic theory are about government debts that are marked to market, while government accounts typically report par values and do a poor job of accounting for coupon payments, let alone default premia. Even the best government accounting systems, such as those of the United States and United Kingdom, report measures of government debt that can deviate markedly from the objects in a macroeconomist’s government budget constraint (for example, see Hall and Sargent 2011).

Institutions delimiting the conduct of monetary and fiscal policy are vital to understanding the origins of inflation and its macroeconomic consequences. Societies are prone to revisit arrangements that set the scope of central bank independence, so these arrangements are naturally subject to controversy and stress. President Trump’s recent public comments about the conduct of monetary policy in the United States are only one example among many instances. Arrangements, rules, and their evolution have contributed to successes in the control of inflation in countries such as Brazil, Chile, Columbia and Paraguay as well as to the failures in Argentina and Venezuela.

The aim of our project has been to compare arrangements and outcomes across countries, to recognize patterns, and to generalize; in doing so we have necessarily ignored some important country specific details and episodes. We strive to describe different regimes using a framework within which policy was conducted. On the other hand, there is scope for deeper probe of the determinants of the different policy regimes in future research. We see this as deliberate and a defensible choice, but acknowledge that it leaves many gaps. Indeed, we see this project as opening the door to investigations that will widen our appreciation of a broader set of long-term and short-term economic and political forces that account for and explain experiences of countries that we have studied, and hopefully other countries too.

While the MFH project offers valuable new data, it does not justify a “just let the data speak for themselves” approach. It is important to use formal models to interpret evidence and reason about the consequences of alternative government policies. While the modeling advances surveyed here are promising, their quantitative and empirical importance remains to be investigated fully. Thus, the chapters in this volume provide information and ideas that promise not only to enhance our understanding of past Latin American experiences but, going forward, to evaluate more generally the relevance of alternative models and to suggest improvements in those models.

Notes

References