Hans Zhu

Over the last decade, U.S. life insurers have aggressively shifted reserves to offshore jurisdictions while simultaneously reallocating portfolios toward private credit. This paper finds that the two are linked through private equity reinsurers domiciled in Bermuda. Insurers benefit directly or indirectly from exposure to higher yielding private credit and the more competitive reinsurance market. Private equity with a large presence in Bermuda can access desired blocks of life insurer capital for their private credit investments as a reinsurer.

Using comprehensive regulatory filings from 2011-2024, we exploit the staggered adoption of offshore reinsurance agreements, to estimate the causal impact of these arrangements on portfolio allocation. Two years following the execution of such a reinsurance agreement with a private-equity backed reinsurer, insurers increase their investments in NAIC 1 long-term bond like private credit by 1.4 percentage points or a third of the median share of investments in private credit.

In addition, these reinsurance and portfolio allocation decisions coincide with changes in the annuity product market. Two years following the execution of such a reinsurance agreement, insurers also increase their annual total annuity premiums (lump sum deposits) collected by around $0.82 billion.

We examine whether policies that ban for-profit providers and allow for subsequent takeovers by not-for-profits can be effective at addressing quality shortfalls. We consider the U.S. nursing home industry, where quality provision has been a concern for many decades. Our motivating evidence suggests that not-for-profit providers choose higher-quality inputs than their for-profit counterparts but are more prevalent in higher socioeconomic markets and serve wealthier residents.

Thus, for-profit providers play an important role in providing access. To explain these facts and explore counterfactual policies, we estimate a structural model of nursing home demand and supply that allows firms to have nonpecuniary motives and costs that differ across provider types. The structural model reveals that for-profit providers have a strong cost advantage for serving needier residents. This results in them choosing a lower-quality and lower-priced product to maximize their margins.

Not-for-profit providers choose higher quality, and therefore higher prices, not because of their nonpecuniary motive, but because they prefer to avoid directly competing with for-profit providers by serving residents with stronger preferences for quality. Therefore, banning for-profit providers while allowing for takeovers tends to reduce consumer surplus because not-for-profits that take over for-profits significantly raise prices to cover their higher costs. These results underscore the role that for-profit providers play in expanding access to nursing home care.

Digital entertainment markets are characterized by many niches and the “long tail” of consumption. I argue that the size of these tails may be dramatically underestimated if ex-post match values—match values realized only after consumption—are not taken into account. I collect a novel individual-level dataset on the Steam platform and interpret playtime data as directly informative about ex-post match values.

The playtime data reveal patterns of consumption starkly different from those inferred from purchases. I then examine the policy relevance of ex-post match values by studying a unique series of policy changes on the Steam platform: the introduction of personalized stores followed by the introduction of refunds. Personalized stores should improve consumer matches on average, but are unable to recommend on the basis of realized ex-post match values.

Refunds allow consumers to realize ex-post match values before committing to the purchase. I find that sales increase following the introduction of refunds by around five times the increase following the introduction of personalized stores. In addition, the sales patterns following the introduction of refunds are dramatically different from those following the introduction of personalized stores. These results suggest that the effects of personalized stores alone are limited. Whether personalized stores matter or are complementary to refunds will be investigated using a structural model.

Complaints of labor shortages are common in health and education markets where prices are often fixed or regulated. In this project, I argue that the two are linked. The ability of providers to adjust their labor inputs depends on their ability to adjust wages, which is limited when they face regulated prices. I study this idea in the U.S. nursing home industry, where a significant proportion of provider revenue comes from fixed prices.

In addition to a structural model of nursing home demand and supply, I model nurse labor supply. This will allow me to consider counterfactuals where nursing homes can freely set prices and study the resulting implications for nursing home quality and the nurse labor market.