Economist @ Center for Economic Studies
     US Census Bureau

Ongoing Projects

Innovation, Productivity Growth and Productivity Dispersion
(with Lucia Foster, Cheryl Grim, John Haltiwanger), A version prepared for the 2017 CRIW can be found here .

The large dispersion in labor productivity across firms within narrowly defined sectors is driven by many factors including, potentially, the underlying innovation dynamics in an industry. One hypothesis is that periods of rapid innovation in products and processes are accompanied by high rates of entry, significant experimentation and, in turn, high paces of reallocation. From this perspective, successful innovators and adopters will grow while unsuccessful innovators will contract and exit. We examine the dynamic relationship between entry, within-industry labor productivity dispersion and within-industry labor productivity growth at the industry level using a new comprehensive firm-level dataset for the U.S. economy. We examine the dynamic relationships using a difference-in-differences analysis including detailed industry moments and focus on differences between High Tech and all other industries. We find a number of distinct patterns. First, we find that a surge of entry within an industry yields an immediate increase in productivity dispersion and then a lagged increase in productivity growth. Second, we find these patterns are more pronounced for the High Tech sector. Third, we find that these patterns change over time suggesting that other forces are at work in the latter part of our sample. We devote considerable attention to discussing the conceptual and measurement challenges for understanding these relationships. Our findings are intended to be exploratory and suggestive of the role innovation plays in the dynamic patterns of entry, productivity dispersion and productivity growth. Given the difficulties in directly measuring innovation, our findings could be used to help identify areas of the economy where innovation may be taking place. Alternatively, our findings suggest a useful cross check for traditional measures of innovation,
Dispersion in Dispersion:  Measuring Establishment-Level Differences in Productivity
(with Lucia Foster, Cheryl Grim, Sabrina Pabilonia, Jay Stewart, and Cindy Zoghi)
Productivity measures are critical for understanding economic performance. The Bureau of Labor Statistics (BLS) produces the official U.S. productivity statistics using aggregate industry level data. However, those statistics cannot provide insight on the within-industry variation in productivity, limiting our understanding of the rich productivity dynamics in the U.S. economy. Existing research has shown that there are large and persistent productivity differences across businesses even within narrowly-defined industries. These differences vary across industries and time and are related to productivity-enhancing reallocation. More generally, dispersion in productivity across businesses is informative about the nature of competition and frictions within sectors. Productivity differences across businesses have also been shown to be related to wage differences across businesses and growth in dispersion is, in turn, related to the rising wage inequality across businesses. To address this gap in productivity statistics, BLS and the Census Bureau are collaborating to create measures of within-industry productivity dispersion with the goal of developing public-use and restricted-use statistics that complement existing statistics. Building on earlier related work, we construct establishment-level labor productivity using Census Bureau microdata. We provide some measure of compatibility by comparing micro-aggregated industry-level measures to BLS industry-level measures of inputs, output and productivity. With some basic summary statistics about these hybrid measures, we begin our exploration of the variation in our industry-level productivity dispersion measures across industries and time.

Using Census Microdata to Forecast U.S. Productivity
(with Eric J. Bartelsman and Christopher J. Kurz), A version prepared for the 2011 NBER CRIW Workshop can be found here .

We contribute to the productivity literature by building a bridge between time series analysis of macro-level productivity growth and firm-level productivity studies. Confirming earlier results that information from firm-level data improves univariate and multivariate forecasts of aggregate productivity (see the 'Published Papers' section on this web page), we use plant-level data from U.S. Manufacturing and extend a state-space model to forecast medium term productivity growth by including a role for heterogeneous firms and the evolution of firm-level productivity and size distributions. We find that adding time series that capture the contribution to productivity growth of within firm productivity growth and reallocation for US manufacturing firms improves estimates and forecasts of trend productivity growth for both the manufacturing and the aggregate non-farm business sectors in the United States. In an updated draft of the paper, we use structural timeseries models to explicitly model  the slowly changing (trend) and cyclical components of aggregate productivity.

Forecastability, Procyclicality and the Granular Hypothesis,
Ph.D. dissertation chapter, 2011.
The dissertation chapter explores the mechanisms that may explain why firm-level information may help forecast aggregate productivity. Using simulation results from a recent theoretical model of heterogenous firms, I show that the productivity components of previous forecast experiments correctly detect shocks to the productivity distribution. This finding helps understand why firm-level information is useful in forecasting aggregate productivity. In addition, I demonstrate that changes in the cyclical behavior of aggregate productivity may be related to changes in the properties of firm-level factor adjustment. To my knowledge, this study is one of the first to demonstrate the relationship between cyclical productivity and factors adjustment costs using a model of firm-level heterogeneity. Finally, I provide evidence that the productivity and market share evolution of large firms may explain aggregate productivity fiuctuations if the firm-size distribution is fat-tailed. This is an important finding because if firm size matters for aggregate developments, then the moments calculated using data on the largest firms may prove useful proxies when data on the entire distribution is not available.
Published Papers

Measuring Productivity Dispersion
(with Eric Bartelsman), Forthcoming, Chapter 18 in The Oxford Handbook of Productivity Analysis

Measuring the dispersion of productivity or efficiency across firms in a market or industry is rife with methodological issues. Nevertheless, the existence of considerable dispersion now is well documented and widely accepted.  Less well understood are the economic features and mechanisms underlying the magnitude of dispersion and how dispersion varies over time or across markets. On the one hand, selection mechanisms in both output and input markets should favor the most productive units through resource reallocation, thereby reducing dispersion. On the other hand, innovation and technological uncertainty tend to increase dispersion. This chapter presents a guide to measurement of dispersion and provides empirical evidence from a selection of countries and industries using a variety of methodologies.

Firm-Level Dispersion in Productivity: Is the Devil in the Details?
(with Lucia Foster, Cheryl Grim and John Haltiwanger),  American Economic Review , May 2016, Vol. 106, No. 5 .  
Download  full paper  and appendix.
We explore current interpretations of firm-level dispersion in revenue-based productivity measures. Since revenue function estimates using proxy methods differ from factor elasticities, the residual emerging from this method remains a combination of demand and technical effciency shocks, and is not equal to the concept of revenue productivity that plays an important role in recent literature on misallocation. This has implications for applications where measured revenue productivity dispersion is used as an indicator of misallocation. Our empirical evidence suggests, under iso-elastic demand, measured dispersion may indicate either distortions or variation in demand shocks and technical effciency or all of the above.

Forecasting Aggregate Productivity using Information from Firm-Level Data
(with Eric J. Bartelsman),  Review of Economics and Statistics, October 2014, Vol. 96, No. 4 , Pages 745-755. 
In this paper, we explore whether information from firm-level data can improve forecasts of aggregate productivity growth. We generate firm-level productivity measures and aggregate them into time-series components that capture within-firm productivity and the productivity contribution of reallocation. We show that these components improve aggregate total factor productivity forecasts in a simple univariate setting, even when firm-level data are available with a time lag. Lagged firm-level information also improves aggregate productivity forecasts when we combine results from a variety of different multivariate forecasting models using Bayesian model averaging techniques.

Working Papers

Macro and Micro Dynamics of Productivity: From Devilish Details to Insights
(with Lucia Foster, Cheryl Grim, and John Haltiwanger), Working Paper 17-41 , Center for Economic Studies, U.S. Census Bureau
(A previous version prepared for 2015 NBER CRIW Workshop can be found  here .)
Researchers have been using a variety of methods to estimate productivity at the firm level. Absent data on prices and quantities, these methods yield what have become known as revenue productivity measures. How these measures are related to physical productivity depends on the assumptions about the environment in which establishments operate. It is perhaps less recognized that the differences across estimation methods have important consequences for interpretation. One such difference concerns revenue function estimates: while cost-share-based coefficients are, in principle, equivalent to factor elasticities, regression-based estimates equal factor elasticities only under strict assumptions about product markets. This implies that revenue residuals are conceptually different under these two broad approaches. Using plant-level manufacturing data for the U.S., we look at the empirical relevance of such distinctions in the context of key stylized facts of the productivity literature. First, we find non-trivial differences in estimated elasticities and returns to scale. The variation in elasticities affects numerical results on dispersion, yet all methods imply large productivity differences across establishments. More productive plants are shown to be more likely to grow and survive by all reviewed methods, although differences remain in the quantitative marginal effects of productivity. Reallocation is found to be comparable and productivity enhancing by all methods considered, but within-plant growth seems to be more sensitive. We find evidence that imputation and imposing homogeneous elasticities negatively affect within-industry dispersion. In addition, imputation results in some attenuation in growth and exit coeffcients but does not invalidate qualitative conclusions.

The Role of Institutions and Firm Heterogeneity for Labour Market Adjustment: Cross-Country Firm-Level Evidence  
(with Alex Hijzen and Peter Gal), IZA Discussion Paper No. 7404 , May 2013.
This project focuses on the role of labor market policies and institutions for aggregate labour market dynamics during the recent nancial crisis. Using firm-level data from a cross-country dataset (ORBIS), we find that differences in firm-level labor adjustment accounts for a nontrivial fraction of the cross-country variation in aggregate employment growth at the outset of the crisis. The framework of our analysis implies we can attribute this to institutional differences. In addition, we offer evidence that stronger employment protection shifts the burden of adjustment from the extensive to the intensive margin of employment adjustment.
Driving factors of growth in Hungary - a decomposition exercise
(with Gabor Katay),  Working Paper 2008/06 , National Bank of Hungary.
Based on a large panel of Hungarian manufacturing firms, we decompose value added growth to input factor, capacity utilization and estimated TFP growth contributions. We find that using an hours worked proxy, the variance of the residual drops considerably. We also find that TFP’s role has not been stable over the period: it contributed to value added growth mostly in periods when/after institutional reforms, privatization or FDI inflow took place and lost its importance several years after the shocks.
Investment Behavior, User Cost and Monetary Policy Transmission - the Case of Hungary
(with Gabor Katay),  Working Paper 2004/12 , National Bank of  of Hungary.
In this paper we investigate corporate investment behavior using a large panel of Hungarian firms between 1993 and 2002. The standard neoclassifical framework is used to derive empirically feasible specifications. We draw on the research carried out in the Eurosystem Monetary Transmission Network (EMTN). Our results are, by and large, similar to those obtained within the EMTN. Namely, the effect of user cost changes on investment is significant and robust across several specifications providing strong evidence against simple sales-accelerator models of investment. Firms' cash-flow proves to be a significant determinant of corporate investment, which suggests that financial position of firms may have consequences for investment.