Mariana Guido

Vertical integration between insurers and primary care practices is increasingly common in the US healthcare market, raising concerns among policymakers. We use medical claims data from the Colorado All Payer Claims Database to study a 2017 insurer-physician acquisition, focusing on how it aligns incentives within the integrated firm. We find two main impacts. First, in Medicare Advantage, where insurer revenue is tied to patients' diagnoses, we leverage a patient-level event study design to estimate how the integrated practice increases the intensity of their diagnostic coding. We find that integration increases diagnosis-based payments to the insurer by $998 to $1,805 per patient per year, without a corresponding increase in treatment. Second, as the integrated firm bears the cost of specialist care, we estimate a referral choice model to evaluate whether the integrated practice steers referrals towards more cost-effective specialists. We find that, for patients in the Commercial segment, the acquired practice steers referrals and saves approximately $300 per inpatient referral. Our results reveal a crucial trade-off: integration can increase taxpayer costs in one market while generating efficiency gains in another. Furthermore, we show that contracting at non-acquired practices can replicate the diagnostic coding effect. This substitutability between integration and contracting suggests that policies targeting only mergers may be ineffective at addressing strategic coding behavior.

Corrective policy in sports betting markets is motivated by concerns that demand is driven by behavioral bias rather than a normative preference to gamble. We conduct a field experiment with frequent sports bettors to measure the impact of two biases, overoptimism about financial returns and self-control problems, on the demand for sports betting. We find widespread over-optimism about financial returns. The average participant predicts that they will break even, but in fact loses about 8 cents for every dollar wagered. Self-control problems are smaller and less common. We estimate a model of biased betting and use it to evaluate several corrective policies. Our estimates imply that the surplus-maximizing corrective excise tax on sports betting is twice as large as prevailing tax rates. We estimate substantial heterogeneity in bias across bettors, which implies that targeted interventions that directly eliminate bias could improve on a tax. However, implementation is challenging: we show that two bias-correction interventions favored by the gambling industry do not deliver the targeting improvements.

PDF survey instruments available for download. Qualtrics files are available on request.

• Survey 1
• Survey 2
• Survey 3

We define the concept of customer overlap of product j with product k as the share of j’s customers who buy k. We then argue that, in appropriate contexts, customer overlaps are an excellent proxy for diversion ratios, a useful and popular way to summarize competition between sellers of substitute products. Unlike diversion ratios, which are often challenging to estimate, customer overlaps are straightforwardly observed in many data sets. We show theoretically, and then validate empirically, the close connection between customer overlaps and diversion ratios. We then illustrate the potential use of customer overlaps in contexts where estimation of diversion ratios could be prohibitive.

Private equity (PE) investment in the U.S. healthcare sector has grown exponentially, raising concerns about its impact on the cost and quality of care. This paper investigates how PE acquisitions alter the behavior of primary care practices (PCPs) and, crucially, explores heterogeneity based on the acquirer's structure. We distinguish between acquisitions by pure financial sponsors and those by PE-insurer joint ventures, which may face conflicting incentives between profit maximization and managing population health costs. We construct a novel national dataset by linking PitchBook data on PE acquisitions from 2006 to 2018 with Medicare fee-for-service claims. Then, we employ a staggered difference-in-differences strategy to compare acquired PCPs to not-yet-acquired PCPs. Using state All-Payer Claims Data (APCD), we will further test for differential effects among Traditional Medicare and Medicare Advantage patients, where insurer-backed owners have stronger incentives to control utilization.

Many people with mental health problems do not seek treatment despite evidence that therapy can improve well-being. Working with a provider of gambling addiction therapy, we conduct a field experiment to measure how behavioral biases impact therapy demand and what interventions may overcome them. We consider two behavioral biases: misperceptions and willpower problems. To study misperceptions, we measure how participants’ perceptions of therapy change after attending a session. To study willpower issues, we measure participants’ willingness to pay for a commitment device to attend therapy. Lastly, we evaluate the effectiveness of two scalable interventions – financial incentives and text message encouragements – at increasing therapy takeup.

Recently implemented policies in the US have mandated employers to provide pay ranges in job applications to combat gender and race gaps. In this paper, we test whether and how such policies impact gender and racial gaps in application decisions and salaries. We conduct randomized experiments and surveys with both applicants and employers. In our experiment with applicants, we study how applicants’ beliefs, application decisions, and ask wages react to the introduction and manipulation of salary ranges on job ads. Through a survey with hiring managers, we provide evidence on how salary ranges are currently constructed, to what extent decision-makers take potential trade-offs into consideration, and whether they have correct beliefs about applicants’ behavior and responses. Status: Fundraised USD 45k+, Piloting

The birth of a child triggers a sharp increase in household chores, disproportionately burdening women and reinforcing persistent gender gaps in labor force participation and earnings (motherhood penalty). While existing policies, such as the provision of subsidized or free childcare and parental leave, aim to support working mothers, they have had limited success in closing the motherhood penalty. In this paper, we conduct an experiment with couples in Brazil to assess the role of intra-household dynamics for the persistence of the motherhood penalty. We test how spouses’ preferences and beliefs over future labor and household responsibilities develop pre- and post-childbirth. Finally, we cross-randomize an information intervention as well as a bargaining intervention, which we co-develop with a startup.