We develop a tractable general equilibrium model of cumulative innovation and growth to study the optimal combinations of three patent instruments: a patentability requirement to protect against future incremental innovation, a patent breadth to protect against current imitation, and the patent length governing the duration of its validity. New ideas strictly build upon frontier technologies, with the size of productivity improvements drawn from a stochastic distribution. The model features positive knowledge spillovers, and we discuss when this leads to under-investment in R&D in the market equilibrium relative to the social planner’s benchmark. We characterize analytically how each patent instrument affects research incentives, and further establish that the patentability requirement and patent breadth in the welfare-maximizing combination are each set at a strictly lower level than in the R&D-maximizing combination. Calibrating the model to aggregate data, the welfare-maximizing combination of instruments features a large patent breadth, a long patent duration, but a modest patentability requirement. The quantitative analysis points to sizeable welfare gains from these patent policies, while highlighting the importance of having multiple patent instruments.
This paper studies the impact of health on structural transformation and aggregate productivity in low-income economies through its effects on the gender-based division of labor across sectors. I develop a two-sector general equilibrium model in which returns to health are higher for men than for women in agriculture, implying that health improvements reallocate female labor toward nonagriculture. The model is calibrated to African economies. Quantitative results show that health improvements substantially reduce the agricultural employment share—particularly among women—and increase aggregate labor productivity. Over one-third of the productivity gains can be attributed to the gender-based division of labor. Policy experiments indicate that health subsidies are more effective than education subsidies in reducing women’s agricultural employment relative to men’s and in improving aggregate productivity and welfare.
This paper examines the aggregate and distributional effects of the Federal Reserve's unconventional monetary policy, comprising quantitative easing and forward guidance, during the 2009-2015 effective lower bound episode, using an estimated medium-scale heterogeneous agent New Keynesian model. Relative to a counterfactual without unconventional policy, these interventions stimulate aggregate activity and generate broad-based welfare gains, with an average gain equivalent to a 0.27 percent increase in lifetime consumption. These welfare gains are mildly U-shaped across the wealth distribution, with households at the bottom gaining mainly from a lower unemployment rate, while higher profits and equity prices benefit the top disproportionately. Comparing these outcomes to an unconstrained conventional interest rate rule, I find that once unconventional tools are available the effective lower bound is not very costly in aggregate terms, while conventional policy alone would have implied somewhat more uneven distributional effects.
This paper examines the quantitative implications of rising college costs, wage inequality, and delinquency for the growth of student debt in the U.S. Rising college costs and wage inequality are introduced as exogenous inputs into an incomplete-markets overlapping-generations model with choices of college attendance, student loans, and delinquency. In the benchmark economy, aggregate student debt rises by $314 billion between 1985 and 2014, accounting for approximately 64% of the observed rise in undergraduate student loans in the U.S. The rise in college costs is the primary driver of the increased borrowing, while the rise in income risk and the decline in student ability lead to higher delinquency rates. Increasing delinquency significantly amplifies debt accumulation: in a counterfactual economy without the option of becoming delinquent, debt increases by only $178 billion. Finally, we show that income-driven repayment (IDR) plans can substantially moderate debt growth, leading to an increase of only $169 billion. This effect arises because IDR reduces delinquency rates by offering greater repayment flexibility.
The Heckman et al. (1998a) (HLT) model includes credit markets and within-period labor supply indivisibilities, two essential features of Ljungqvist and Sargent (2006) “time-averaging” models. But by assuming inelastic labor supplies until a mandatory retirement age, it shuts down time-averaging. We activate time-averaging by endogenizing retirement ages. Our addition of a baseline social security system puts all workers at corner solutions of their retirement decisions, letting our model reproduce most outcomes in HLT's model. By dislodging workers from those corners, social security and tax reforms raise the aggregate labor supply elasticity and can bring about a “dual labor market.” HLT's Ben-Porath human capital technologies generate steeper earnings profiles for college-educated workers that in our model make their labor supplies more resilient to tax and social security reforms than high school workers' labor supplies. But nonconvexities inherent in the Ben-Porath technologies can bring “tipping points” at which tax increases cause workers who at lower tax rates had chosen long careers and made substantial human capital investments to jump discretely to choosing much shorter careers and doing much less on-the-job human capital accumulation.
Theories are ambiguous about whether households of higher or lower wealth consistently trade against expected returns. I employ unique matched data on US household trades in the housing market and their wealth to study trade timings and quantify the implications for portfolio returns and wealth inequality. I find that poorer households consistently buy houses when expected returns are low (e.g., after price increases) and sell when expected returns are high (e.g., after price declines). The estimated dispersion in expected portfolio returns from active trades is large, with an interquartile range across the wealth distribution of 65 basis points per year. Back-of-envelope calculations suggest this return dispersion could account for roughly one-fifth of observed wealth concentration beyond what income differences alone explain, within the interquartile range.
This paper studies how labor market power affects firm dynamics and aggregate productivity. We build a dynamic model of neoclassical monopsony with occupational choice, firm growth, and productivity-enhancing technology adoption. Labor market power lowers efficiency and leads to aggregate output losses by distorting the allocation of labor, entrepreneurship, and innovation decisions. The model is consistent with cross-country evidence of higher life cycle firm growth and higher productivity investment in more competitive labor markets and can explain 25 percent of the differences in aggregate productivity across countries. We find that about 85 percent of the losses are attributable to the lack of technology adoption induced by weaker labor market competition, suggesting that efficiency losses may be greater than those estimated by previous studies.
Using transaction-level data from the Korean tri-party repo market, we study how repo contract terms interact with collateral quality. Both haircuts and interest rates rise with collateral risk, consistent with existing evidence. Conditional on collateral quality, however, we find a trade-off: a one-percentage-point increase in the spread is associated with a 1.3-percentage-point reduction in the haircut. The same increase in the interest rate is associated with a smaller reduction in the haircut under heightened market uncertainty, indicating that insurance against default becomes more valuable as default risk rises. We show that the interaction between lenders’ incentives to acquire information and borrowers’ opportunistic default risk explains both the positive unconditional relationship and the negative conditional relationship between haircuts and interest rates.
Student debt decreases post-bachelor school enrollment and earnings growth but does not delay first-time home ownership. We introduce a life-cycle human capital model with wealth heterogeneity and financial frictions and show that high debt balances distort career choices because returns to further education depend on current income. Student debt impacts home ownership in two ways. First, it deters ownership via the traditional wealth channel. Second, it increases ownership by discouraging further education in favor of early labor market entry. Finally, we discuss the impact of student borrowing under different loan repayment schemes.
This paper studies the distributional outcomes of protectionism. First, we investigate the short-run distributional effects on workers with different skill levels by estimating structural vector autoregressions (VARs) using high-frequency measures of temporary trade barriers for the United States. We then estimate a panel VAR for a sample of thirty-six countries using the applied tariff rates. Across our empirical exercises, we find robust evidence that protectionism reduces the skill premium but increases the employment ratio between high-skilled and low-skilled workers. To rationalize these findings, we build a two-country dynamic general equilibrium model featuring asymmetric search-and-matching (SAM) frictions, capital-skill complementarity (CSC) in production, and producer dynamics. Our model results replicate the empirical patterns. Our counterfactual analysis highlights the interaction between asymmetric SAM and CSC in qualitatively shaping the distributional patterns of protectionism, with producer dynamics magnifying these effects quantitatively.
I show that firms' ability to postpone entry has important implications for our understanding of the observed business cycle behavior of start-ups. I use a model that closely replicates the main features of the US firm dynamics to explore and quantify the mechanism. I find that the option to wait endogenously generates a countercyclical opportunity cost of entry: during recessions, a higher risk of failure increases the value of waiting, hence the cost of entry. The mechanism increases the elasticity of entrants to aggregate shocks five times. It is responsible for three-fourths of the observed persistent differences in the recessionary and expansionary cohorts' productivity, survival, and employment. Without the channel, existing models require either large shocks that generate excessive aggregate fluctuations or exogenous mechanisms to reconcile the observed dynamics of entrants. Overlooking this channel may also result in misleading predictions about entrants' responses to different shocks or policies.
We estimate a tractable income process that is consistent with key facts on individual income risk and its variation over the business cycle. In particular, the estimated process generates income fluctuations that display (i) flat and acyclical variance, (ii) volatile and procyclical skewness, (iii) very high kurtosis, and (iv) a moderate rise in within-cohort inequality over the life cycle, all consistent with the US data. Furthermore, the income process captures the predictable nature of business cycle income risk: income changes during a business cycle episode are partly predicted by income levels before that episode. The estimated process features a time-varying distribution of innovations as well as a factor structure for business cycle exposure. Incorporating the estimated process into a business cycle model adds only one state variable—as in the workhorse persistent-plus-transitory income process—making it a tractable option for modelers.
A long-standing question in health economics asks what drives heterogeneity in health investments across sociodemographic groups, which contribute to health disparities. We examine a novel source: variation by sociodemographic groups in the impact of medical treatment side effects. We estimate a lifecycle model of medication and labor supply decisions using data on men infected with human immunodeficiency virus (HIV) and with different levels of completed education. Agents in the model make dynamic medication and work choices as a function of expected income, health, and mortality. We also include utility costs of side effects, which can interact with the utility cost of work, and allow these parameters to vary by education. We use the estimated model to evaluate the disparate impacts of an effective HIV treatment innovation that had harsh side effects: HAART (which stands for highly active antiretroviral treatment). Measured in lifetime utility gains, HAART disproportionately benefited patients with more education in part due to differences in income and mortality, but also due to smaller impacts of side effects on labor supply. We also simulate the effects of a HAART treatment mandate, which mimics assignment to treatment in a clinical trial. The mandate improves health, which might be viewed as a success in a randomized trial. However, this masks the downsides of treatment, including lower labor supply due to side effects, especially for lower-education men. Broadly, viewing heterogeneity in health investments as solely a result of barriers to access is overly simplistic. Individually optimal investments in health may in part reflect the costs of managing side effects and work. If low health investments create negative externalities (e.g., due to increased use of publicly-funded healthcare), the benefits of policies that moderate resulting health-work tradeoffs could outweigh the costs.
We develop a general equilibrium model of the U.S. mortgage market where securitization shapes households' access to credit and housing. The model matches the long-run behavior of mortgage rates, credit growth, and house prices following the development of securitization. Securitization improves market efficiency but is subject to adverse selection: lenders are privately informed about mortgage quality. Surges in household defaults then drive down security prices, reducing lender liquidity and contracting credit supply. Applied to the Global Financial Crisis (GFC), the model replicates two-thirds of the contraction in residential mortgage credit observed in the data. How important were information frictions in accounting for these dynamics? We estimate an information-friction multiplier of around 1.2, suggesting a powerful amplification effect. Evaluating post-GFC credit guarantee policies, we find that expanded coverage has stabilized credit but guarantees remain underpriced. Pricing them to reflect the amplification effect of information frictions covers the associated fiscal costs and delivers welfare gains for all households.
This paper estimates two measures of human capital externalities. By incorporating externalities into an overlapping-generations model of human capital accumulation with Compulsory Schooling Laws (CSL), we show that human capital externalities can be estimated from the effects of CSL for one generation on wages of other generations. Using an instrumental-variable strategy deduced from the model, we find one more year of average schooling at the U.S. state level raises individual wages by 6-8%. Taking this reduced-form estimate into account, we find the elasticity of a typical firm’s productivity with respect to the average human capital of an economy is 0.121.
This paper develops a competitive search model with asymmetric information and argues that small firms hold liquid assets not only to self-finance their investments but also to signal their investment quality, thereby obtaining better loan terms. Because self-finance is an outside option of borrowing, it affects bank loans and the credit market structure. Monetary policy affects the cost of self-finance and hence affects the market structure and the screening regimes in the credit market. An increase in the policy rate can trigger banks to use screening contracts, which distort allocations. Using U.S. data, I document that dispersion in firms' cash-to-asset ratios is hump-shaped in policy rates, consistent with the model's prediction that firms use liquidity holdings to signal.
I estimate CES aggregate production functions for the US, the UK, Japan, Germany, and Spain using data from the EU KLEMS database. I distinguish between three types of capital: information and communication technologies (ICT), intellectual property (IP) capital, and traditional capital. I assume that the aggregate output is produced using labor and these three types of capital and allow for differences in the elasticities of substitution between labor, an aggregate of ICT and IP capital, and traditional capital. The estimated elasticities of substitution between ICT and IP capital are strictly below one for all sample countries implying gross complementarity. ICT and IP capital together are gross substitutes for labor, while traditional capital is a gross complement. The results for the US imply that the fast pace of technological progress in ICT and IP capital accumulation together are responsible for about 80 percent of the fall in labor income share.
We use a unique dataset and novel empirical strategy to validate the predictions of labor supply models featuring returns to experience, quantify the bias in standard estimates of the intertemporal elasticity of substitution (IES) that assume exogenous wage formation, and obtain an unbiased estimate of the IES. Our approach uses the insight that the bias in standard estimates shrinks as returns to experience decline in importance relative to the wage. Our identification strategy does not rely on the structure of the human capital accumulation process but does require observing workers over multiple periods at the very end of their careers. Using data on the daily labor supply of Florida fishermen, we obtain an estimate of the IES that is large (2.5). Consistent with the theory, we find that a "naive" estimate is biased downward by nearly a factor of 2 and that the bias is larger for fishermen early in their career. (JEL E24, J22, J24, J31)
We provide a theorem on the role of risk and risk attitudes in macroeconomic models that clarifies and extends the Tallarini (2000) separation result. Under (1) separation of intertemporal and risk preferences, (2) separation of drivers of first and higher moments in the model primitives, and (3) approximate linearity of constraints, risk aversion and time-varying risk are irrelevant for the elasticity of any endogenous variable with respect to state variables that don't drive variation in higher moments. We discuss how models generate a more prominent role for risk by “breaking” or “adapting” to the assumptions in the theorem.
