Structural DSGE models are used both for analyzing policy and the sources of business cycles. Conclusions based on full structural models are, however, potentially affected by misspecification. A competing method is to use partially identified VARs based on narrative shocks. This paper asks whether both approaches agree. First, I show that, theoretically, the narrative VAR approach is valid in a class of DSGE models with Taylor-type policy rules. Second, I quantify whether the two approaches also agree empirically, that is whether DSGE model restrictions on the VARs and the narrative variables are supported by the data. To that end, I first adapt the existing methods for shock identification with external instruments for Bayesian VARs in the SUR framework. I also extend the DSGE-VAR framework to incorporate these instruments. Based on a standard DSGE model with fiscal rules, my results indicate that the DSGE model identification is at odds with the narrative information as measured by the marginal likelihood. I trace this discrepancy to differences both in impulse responses and identified historical shocks.

(forthcoming, Review of Economic Dynamics)

We quantify the fiscal multipliers in response to the American Recovery and Reinvestment Act (ARRA) of 2009. We extend the benchmark Smets-Wouters (Smets and Wouters, 2007) New Keynesian model, allowing for credit-constrained households, the zero lower bound, government capital and distortionary taxation. The posterior yields modestly positive short-run multipliers around 0.52 and modestly negative long-run multipliers around -0.42. The multiplier is sensitive to the fraction of transfers given to credit-constrained households, the duration of the zero lower bound and the capital. The stimulus results in negative welfare effects for unconstrained agents. The constrained agents gain, if they discount the future substantially.

Paper (final version coming soon)
Column on

Entrepreneurial Tail Risk: Implications for Employment Dynamics

Starting a new business is risky. This paper asks how entrepreneurial productivity risk has changed over time in the US and how this has affected employment. I estimate this risk using micro data on the size distribution of new businesses and their exit rates. The paper distinguishes upside and downside risk and analyzes their effects in a tractable dynamic general equilibrium model of entrepreneurship. At the heart of the model is the decision of potential entrepreneurs whether to start a new business in the face of uninsurable productivity risk. Applied to US time series data, structural estimates suggest that higher upside risk explains much of the high job creation in the late 1990s. Over the entire sample period, time variation in risk explains 40-55% of the variation in employment of new businesses. Reduced form results show that this relationship is strongest in IT-related industries. Counterfactual simulations show that the explanatory power of a model with a single risk factor drops by 30-50% compared to the baseline estimates.

Paper  (updated August 2014)

Work in Progress

Accounting for the Sources of Macroeconomic Tail Risks (with Enghin Atalay)

We empirically examine the sources of aggregate tail risks. Using data on sectoral employment from 1947 to the present, in conjunction with a multi-industry model of the type introduced in Long and Plosser (1983), we calculate the contribution of industries' disturbances to aggregate output and employment tail risks. We find that a pure statistical model of industry-level employment growth would misrepresent the importance of shocks in different industries. In ongoing work we expand our model to allow for industry-specific capital and consumer durables.

All jobs are created equal? Examining the importance of startups for local labor demand (with Jerry Carlino)

We investigate the dynamic response of local US labor markets to increased job creation by new businesses and compare the effects to overall labor demand shocks. To account for both dynamic and spatial dependence we employ a panel VAR that allows for varying degrees of spatial dependence. Following recent advances in the VAR literature we identify structural shocks using predictors based on historical industry structure and national industry-specific labor demand and job creation as instruments. We find that shocks to job creation by startups explain twice as much of the local variation in population growth than shocks to overall labor demand and about 70% of the variation in wage growth. We also find evidence of spillovers from jobs created by startups to employment at incumbents. This is consistent with existing studies on the spillover effects of the entry of new plants. Taking account of the spatial dependence amplifies the effects of these shocks on impact by about 3% for population growth and about 25% for exit rates for a typical location.