Final Report Summary - HUMP (Heterogeneity, Uncertainty and Macroeconomic Performance)
The HUMP project looks at the interaction of risk, heterogeneity of households and firms, and macroeconomic outcomes. Labor market risks are the key theme in the project – in particular movements into and out of unemployment as well as between jobs.
These labor market risks are important for the macro-economy for two key reasons. On the upside, movements of production factors between production units are pivotal for the overall efficiency of the economy. As a downside, they create substantial income risks for households. If households cannot fully and insure against these risks on insurance markets, households will do so implicitly using savings in assets such as government bonds, and stocks, but also houses and other fixed assets. Importantly, this implies that whenever the risks change households face, their propensity to consume changes as well. This, as swings in aggregate demand, can create feedbacks into the macro-economy.
The project has made substantial advances in understanding the latter aspects, both empirically and theoretically: Empirically, we documented self-insurance mechanisms being not only reflected in data on demand for consumption goods but also in data on life-satisfaction. Theoretically, we showed that realistic changes in household-income uncertainty lead to a significant decline in aggregate activity. Upon an increase in uncertainty, households save more, but invest particularly in liquid, nominal assets well suited for short-term consumption smoothing. They even decrease their demand for illiquid, real, assets. Therefore, consumption and investment demand contemporaneously falls. Quantitatively, we show that changes in the uncertainty of household income operating through this channel are roughly twice as important as changes in the uncertainty about aggregate economic policy.
With respect to understanding the dynamic allocation of production factors, the project has made advances in making available and analyzing a new data base of quarterly job and worker flows for the universe of German establishments since 1975. We documented the main cyclical characteristics: Job creation falls during recessions, job destruction shows a short-lived spike early in recessions but is acyclical otherwise. Worker flows in excess of job flows are strongly procyclical, implying that booms are times where more workers reallocate across plants. This large worker sorting during booms is not coming from sorting to plants with particular observables as large, old and high paying plants face both higher accession and separation rates of workers during booms. Likely booms are times where match quality between workers and establishments becomes better.
Moreover, we have documented differences and similarities in job and worker flows in East- and West-Germany to understand the dramatic, persistent difference in labor productivity between the two regions. While East-Germany is 30 % less productive, we find that this does not relate to low productive plants being kept alive by subsidies, to lower worker skills, to industry composition, or to differences in the physical capital stock. Even the plant-size distribution in the East mimics the one in the West – just as if generated for the West-German productivity distribution, but with 30 % lower average productivity. What differs crucially is that plant survival is shorter and job and worker reallocation is higher in the East. We argue that it is likely the low rate of job survival that kills incentives to invest search effort into the quality of worker-job matches which is responsible for low aggregate productivity.
Finally, we have documented and linked dispersions of factor productivities within Germany, Colombia, Chile and Indonesia to a model of technology choice. We find that most of the dispersion stems from lasting productivity differences that are negatively correlated across factors. Put simply, some plants operate a more labor intensive technology than others. We show that these productivity dispersions are rather an epiphenomenon of growth than a cause of low aggregate productivity.
These labor market risks are important for the macro-economy for two key reasons. On the upside, movements of production factors between production units are pivotal for the overall efficiency of the economy. As a downside, they create substantial income risks for households. If households cannot fully and insure against these risks on insurance markets, households will do so implicitly using savings in assets such as government bonds, and stocks, but also houses and other fixed assets. Importantly, this implies that whenever the risks change households face, their propensity to consume changes as well. This, as swings in aggregate demand, can create feedbacks into the macro-economy.
The project has made substantial advances in understanding the latter aspects, both empirically and theoretically: Empirically, we documented self-insurance mechanisms being not only reflected in data on demand for consumption goods but also in data on life-satisfaction. Theoretically, we showed that realistic changes in household-income uncertainty lead to a significant decline in aggregate activity. Upon an increase in uncertainty, households save more, but invest particularly in liquid, nominal assets well suited for short-term consumption smoothing. They even decrease their demand for illiquid, real, assets. Therefore, consumption and investment demand contemporaneously falls. Quantitatively, we show that changes in the uncertainty of household income operating through this channel are roughly twice as important as changes in the uncertainty about aggregate economic policy.
With respect to understanding the dynamic allocation of production factors, the project has made advances in making available and analyzing a new data base of quarterly job and worker flows for the universe of German establishments since 1975. We documented the main cyclical characteristics: Job creation falls during recessions, job destruction shows a short-lived spike early in recessions but is acyclical otherwise. Worker flows in excess of job flows are strongly procyclical, implying that booms are times where more workers reallocate across plants. This large worker sorting during booms is not coming from sorting to plants with particular observables as large, old and high paying plants face both higher accession and separation rates of workers during booms. Likely booms are times where match quality between workers and establishments becomes better.
Moreover, we have documented differences and similarities in job and worker flows in East- and West-Germany to understand the dramatic, persistent difference in labor productivity between the two regions. While East-Germany is 30 % less productive, we find that this does not relate to low productive plants being kept alive by subsidies, to lower worker skills, to industry composition, or to differences in the physical capital stock. Even the plant-size distribution in the East mimics the one in the West – just as if generated for the West-German productivity distribution, but with 30 % lower average productivity. What differs crucially is that plant survival is shorter and job and worker reallocation is higher in the East. We argue that it is likely the low rate of job survival that kills incentives to invest search effort into the quality of worker-job matches which is responsible for low aggregate productivity.
Finally, we have documented and linked dispersions of factor productivities within Germany, Colombia, Chile and Indonesia to a model of technology choice. We find that most of the dispersion stems from lasting productivity differences that are negatively correlated across factors. Put simply, some plants operate a more labor intensive technology than others. We show that these productivity dispersions are rather an epiphenomenon of growth than a cause of low aggregate productivity.