The Stabilizing Effect of Referral-Networks on the Labor Market
Abstract: How does job search using informal connections (referral networks) affect the severity and length of recessions? Moreover, are policies designed to affect the use of referral-networks more effective than traditional methods in hastening post-recession recovery? This paper answers these questions using a search-and matching model in which there are two hiring methods, formal channels and informal channels, and workers endogenously adjust their network of informal contacts in response to shocks and government policy. We show referral-networks have a stabilizing effect on the labor market, reducing the severity of adverse economic shocks and accelerating post-recession recovery. Counterfactuals demonstrate the government must exercise caution when enacting policies intended to expedite economic recovery. Policies that generically improve worker-firm matching prolong recovery by 8 months, as they facilitate relatively more matches between workers and low-productivity firms during recessions. In contrast, policies aimed at reducing the costs of network-formation or increasing referral-network prevalence facilitate more matches between workers and high-productivity firms, expediting recovery by 3-6 months. Slides (8/21/2019)
The Impact of Referrals on Sectoral Reallocation
Abstract: This paper investigates a new explanation for the long-run decline in sectoral switching--the increased prevalence of referral-networks. Using data from the Current Population Survey (CPS), I first document empirically significant increase in the use of referral-networks by the unemployed that is concurrent with the decline in sectoral switching. The CPS data are then used to estimate the effect of using referral-networks on the likelihood of an individual switching sectors. For all industry aggregations, using referral-networks significantly reduces the probability a worker switches sectors. After controlling for demographics, these estimates imply an increase in referral-network use could explain as much as 5% to 40% of the decline in sectoral switching. To better illustrate the policy implications of this finding, a discrete time sectoral-switching model is constructed using a search and matching framework with labor market referrals. The estimated model predicts a referral-switching elasticity of about -.12, which is within the empirically estimated range of -.05 to -.22 for the 2-sector industry aggregation, demonstrating that the increased of the prevalence of referrals overtime can explain about 20\% of the decline in US sectoral switching. Welfare experiments indicate referrals are a ``benign'' cause of the decline, i.e. welfare declines upon effectively banning the use of referral-networks. These results suggest the cause of the decline in sectoral switching is the result of improved matching efficiency over time rather than market inefficiency.
Does Job-Finding Using Informal Connections Reduce Mismatch? The Role of Nonpecuniary Benefits
Abstract: This paper presents evidence that nonpecuniary benefits of a job, such as hours, commute time, and work environment, are a salient factor in a worker's decision to either accept or reject the offer. Using data form the Survey of Consumer Expectations (SCE), I document three empirical facts on the use of referral-networks and mismatch. First, not all referrals reduce perceived mismatch as reported by workers. For high-skill workers, referrals from former coworkers tend to reduce perceived nonpecuniary-mismatch. For low-skill workers, referrals from friends and family tend to increase perceived non-pecuniary mismatch. Given these empirical facts, I construct a search-and-matching model of the labor market similar to Buhrmann [2018a] where workers and firms are given types on a unit interval and suffer increasingly greater productivity losses depending on distance between the firm's type and the worker's type. I augment this baseline model with mismatch along two dimensions -- skill and nonpecuniary preferences-- and calibrate it to the US economy. Results show nonpecuniary preferences can generate more dispersion in skill-mismatch for very low-skill workers and very high-skill workers. Moreover, while referral-networks generally improve aggregate mismatch, they have a heterogeneous affect on nonpecuniary mismatch by type. For low-skill (high-skill) workers, referral-networks increase (decrease) nonpecuniary mismatch.
Broken Instruments (with Trevor Gallen)
Repeated use of the same instrumental variable by a literature can “collectively invalidate" an instrument. This paper examines two ways in which this can happen. First, when the same instrumental variable is used to instrument multiple distinct covariates, it is more likely to violate the exclusion restriction. Second, when a variable is documented to affect many outcomes that are likely to be highly or even mildly persistent, using lagged values of that variable as an instrument is likely to violate the exclusion condition. This paper produces a dataset of approximately 960 instrumental variables papers from 1995-2019 in highly-ranked economics general interest and field journals. We find six commonly-used instruments whose literatures, taken together, suggest they are likely to fail the strict exogeneity condition: (i) elevation and bodies of water (ii) sibling structure (iii) ethnicity/ethnolinguistic fractionalization (iv) religion (v) weather and (vi) immigrant enclaves. Taken together, these instruments have been used in 86 “top five" publications and 317 well-ranked field or general interest journals, with 189 total uses cataloged from 2011 onwards. We conduct Monte Carlo exercises and suggest methods to determine whether or not an IV regression’s point estimates are likely to be correct.