Essays on Secular Stagnation in the USA

The dissertation is about Secular Stagnation in the United States and consists of three papers. The concept of Secular Stagnation was introduced with the pioneering work by Alvin Hansen (1939) to describe the somber situation in which the US economy fell after the Great Recession in 1929. The author looked at the high unemployment as the principal problem for Americans and that expression stood for “sick recoveries which die in their infancy and depressions which feed on themselves and leave a hard and seemingly immovable core of unemployment”. While dismissed in subsequent periods, that concept bloomed again in conjunction of the Great Recession in 2007: Larry Summers (2014), indeed, recalls it to outline a situation in which changes in the economic fundamentals might have brought about a shift in the natural balance between savings and investments. The corresponding natural rate of interest associated with full employment would then have decreased toward negative values, making monetary policy ineffective. The relative outcome would feed a state-of-affairs that makes the attainment of adequate growth, capacity utilization and financial stability very arduous. In the first paper, entitled “Does the Secular Stagnation hypothesis match with data? Evidence from USA”, I decide to take a historical perspective to see which characteristics associated to Secular Stagnation are found in the data. The paper adds to the debate in four ways. Precisely, I focus my study on US macroeconomic data about real per capita GDP, potential output, productivity measures and population since 1870, when possible. This very simple setting allows me to prove that the slow growth in real per capita GDP as in more recent times should not be interpreted as an evidence of Secular Stagnation. Rather, it represents the return to normal, that is the average growth rate experienced before the Golden Age period 1950-1972. In contrast, it is apt to talk about Secular Stagnation in terms of productivity growth (labour and multifactor), since their decline is greater than any previous shortfall. My findings cast some doubt on Summers’s hypothesis of negative natural rates, which suffers from theoretical inconsistencies too. In contrast, a careful analysis of data offers some evidence supporting to Gordon (2015) and Hein (2016a)’s Secular Stagnation hypotheses, among others. Moreover, this evidence shows that the use of the term “Secular Stagnation” in the literature is somewhat misleading, since it should concern a longer time span, possibly involving more extended long runs. Finally, albeit the great heterogeneity in the approaches implemented, I trace out a complementarity or even convergence to what policymakers should do to get away from this trap. In the second paper, entitled “Secular Stagnation and innovation dynamics: An agent-based SFC model. Part I”, I start by noticing that the debate on Secular Stagnation paid little attention to the deep relationship between income distribution, innovation and productivity. The paper fills that gap in the literature and sets Secular Stagnation into an agent-based framework. I focus on the US capitalistic evolution of last fifty years and study in which way the distribution of income between wages and profits can determine the rate of innovative activity and then further attainments in productivity. I consider major features of the US post-1972 economy like the progressive worsening of the functional distribution of income at the expense of the labour share and, on the other hand, a slower growth in R&D activity. In this contribution, I develop an agent-based SFC model in the line of Dosi et al. (2010), Caiani et al. (2016) and Godley-Lavoie (2006). In my model, the micro-foundation of the endogenous innovation process is essential to avoid isomorphism between micro and macro phenomena and to remark the evolutionary character of the theoretical base. Additionally, agent-based models are particularly suitable to the task since the user knows by construction the micro data-generating process and can explore the features of macro-variables as properties emerging out of the evolutionary dynamics. The contribution of the paper to the literature on Secular Stagnation lies on its capability to show the way phenomena at the macro-level affect the dynamic path of variables at the micro level. Precisely, it is interesting to show that the increase of income inequality at the expense of the labour share impacts negatively on firm’s propensity and ability to innovate. I advance the idea that the continuous shift of income from wages to profits may have resulted in a smaller incentive to invest in R&D activity, entailing the evident decline in productivity performances that marks the US Secular Stagnation. I must admit, of course, that this is not the only valid explanation for the long-run tendency of productivity growth to fall. Non-technological motives, like lower top marginal tax rates, increased low-skill immigration, rising trade with China and low-cost manufacturing countries or the rise of superstar firms are equally important. However, the model in the second paper does not deal with growth question but analyzes economic systems that gravitate and fluctuate around a stationary state. In the third paper, entitled “Secular Stagnation and innovation dynamics: An agent-based SFC model. Part II”, I extend and complete the argument started with the second paper. In other words, I develop an agent-based, stock-flow consistent model to analyze the nexus between income distribution and innovative search in determining economic growth. The model is complex, adaptive and structural in the sense of Tesfatsion (2006) and Dosi et al. (2010). First, it is complex because the system is composed of interacting units. Second, it is adaptive since involves environmental changes. And third, it is structural because it builds on a representation of what agents do. In this context, agent are encapsulated set of data and behaviors representing an entity residing in a computationally constructed world. The model manages to replicate several well-established stylized facts of the literature. More precisely and observing the microeconomic level, firms exhibit skewness and heavy tailed-ness in their size distribution, and they are highly heterogeneous in terms of productivity. Moreover, investment lumpiness is an interesting outcome of the model. At the same time, the model respects some empirical regularity at the macroeconomic level, such as endogenous and self-sustained economic growth, fluctuations at the business-cycle frequencies, and volatility and correlation patterns between macro-variables. Theoretical policy implications do not change significantly from the second paper for what regards to distributive policies and their relationship with innovation rates. However, they do change for what concerns the role of the interest rate. What I grasp is that the interest rate has a non-linear and small effect upon innovation efforts and on the overall level of economic activity. More precisely, the very non-linearity in the R&D pattern arises because of the contrasting movement between the revenue and the cost components. On the one hand, capitalists increase the consumption in absolute terms because more profits accrue to their pockets and their need to innovate rises; but on the other hand, they are less afraid of the competitive pressure and reach a normal profit rate more easily, so the necessity to seek for labor-saving techniques looks reduced. Last step of the paper is about econometrics. I want indeed to test the predictions from our model to the empirical level. In so doing, I gather a panel of US manufacturing industries with data on total R&D expenditure, hourly wage rates, productivity levels and values of shipments from 1953 to 2011. I carry out a twofold empirical analysis. First, I try to find empirical evidence of a positive and long-run relationship between R&D spending and its revenue and cost components. The latter are identified, respectively, with sales or shipments and productivity-adjusted wages. I can detect these positive and long-lasting evidence, confirming the predictions of my ACE model. Moreover, the robustness of my results is assessed through different econometric methods applied to panel data. Secondly, I do a simple descriptive analysis to observe the time-evolution of R&D and interest rates, finding that a well-behaved and straightforward interplay between them probably does not exist. Still, that does not conflict with my expectations.