Publications

This paper analyzes the firm’s choice between subsidy support and loan support during the COVID-19 crisis and explores the implications of this choice on firms’ employment growth. We compile a novel micro-level dataset of Lithuanian firms’ balance sheet data and government support records during the pandemic period. We use the dataset to provide a set of stylized facts, categorizing the variety of enacted support policies and tracking aid distribution patterns. We show that larger firms were more likely to choose loans over subsidies. This result cannot be fully explained by policy eligibility criteria and the severity of the pandemic shock, suggesting that firm characteristics played a significant role. Finally, we show that the type of support has implications for firms’ outcomes – subsidy-recipient firms experienced higher employment growth compared to loan-recipient firms.

We show that advance information on future income can be identified from the correlation between consumption growth and future income growth conditional on current income growth. Employing PSID data, we find that this conditional correlation is positive and significant. We use this evidence to structurally estimate a standard incomplete markets model and discover that US households possess enough advance information to reduce their income forecast errors by 15%. This significantly affects the measurement of consumption insurance. With advance information, 25% more income shocks pass through to consumption on average, and more than twice as much for the 5% asset poorest.

Keywords: income risk, advance information, consumption insurance, panel data, incomplete markets.

JEL Classification: C23, D12, D31, D52, D81, E21, G52.

This paper studies the effect of school closure on student outcomes in the Lithuanian context. Using administrative student-level data over 2013–2017 and propensity score matching, we create a balanced sample of control and treatment groups. In contrast to other studies, we focus on students in the final years of high school, possibly eliciting the upper bar of the disruption effect. Also, we follow students after high school graduation, providing evidence on labor market outcomes. We find that the school closure effect depends on the main teaching language. If we match students on a large set of student and school characteristics but the main teaching language, school closings have a lasting negative effect on exam performance and enrolling in higher education. Matching students on the main teaching language significantly reduces the negative school closure effect, suggesting that the disruption effect is considerably smaller and also has limited outcomes after high school if we take the main teaching language into account.

Keywords: School closure, education finance, student outcomes.
JEL Classification: H52, I22, I24.

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.

Subprime lending is concentrated among minorities and in minority neighborhoods. However, the literature has little evidence for what led to the rise of subprime lending in minority neighborhoods. We use the endorsement of FICO credit scores in mortgage underwriting by the Government Sponsored Enterprises (GSEs) in 1995 to answer this question. The use of credit scores led to the sorting of prime and subprime lenders across minority and non-minority neighborhoods. In minority neighborhoods prime lenders were substituted by subprime lenders and, as a result, the share of subprime lending in minority neighborhoods increased by 5 percentage points. Prime lenders with a stronger relationship with the GSEs reduced their lending in minority neighborhoods more. The level of securitization by the GSEs in minority neighborhoods also decreased.

Keywords: Mortgages, Subprime lenders, GSEs, Securitization, Minorities

JEL codes: G21, G28, J15, R23.

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.

Using the universe of mortgage applications data and detailed credit performance data, we document that since the early 1990s there was a significant decline in the share of mortgages with co-borrowers. Although the decline was an almost universal phenomenon across different regions of the US, the rate of the decline showed significant spatial heterogeneity and in turn had implications for regional differences in economic activity. We show that the presence of a co-borrower reduces the mortgage default probability by more than 50 percent for both prime and subprime loans and those regions that had a lower co-borrower share prior to the crisis experienced higher mortgage default rates over the period 2007-2010. Higher default rates created spillovers on economic activity during the Great Recession: a lower co-borrower share at the regional level was also related to persistently lower house price growth, refinancing growth and mortgage credit growth. These results imply that the decrease in the share of mortgages with co-borrowers made the US mortgage market more vulnerable to the financial crisis and contributed to the divergence in economic outcomes across different regions.

JEL Codes: G21, G51, R21. 

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.

The objective of increased regulation of mortgage origination activities after the Great Recession was to prevent another foreclosure crisis in the future. However, the literature is not conclusive about the actual effect of these policy changes. By using the 2007-09 panel and subsequent waves of the Survey of Consumer Finances (SCF), we predict foreclosure risk based on individual borrower characteristics. We show that the median mortgage foreclosure probability kept decreasing after 2010, but in 2016 it was still higher relative to the year 2007. The median foreclosure probability has remained high among both non-bank borrowers and bank borrowers. The regulatory changes started in 2010, so we also compare predicted foreclosure probabilities to the levels in 2010 and find that, despite the fact that banks were affected by this regulation more than non-banks, predicted foreclosure probabilities for bank mortgages declined slower than for non-bank mortgages. Our findings offer support for a thorough analysis of the regulatory effects because they might have been weaker than expected or worked in an unexpected way.

JEL Codes: C53, G21, G23.

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.

We develop a consumption-savings model that distinguishes households’ perceived income uncertainty from income uncertainty as measured by an econometrician. Households receive signals on their future disposable income that can drive a gap between the two uncertainties. With an uncertainty gap that is consistent with direct estimates stemming from subjective income expectations, the model jointly explains three consumption inequality and insurance measures in US micro data that are not captured without the difference: (i) the cross-sectional variance of households’ consumption, (ii) the covariance of current consumption and income growth and (iii) the income-conditional mean of household consumption.

JEL Codes: E21, D31, D52.

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania. 

Using the March Current Population Survey, I show that over the last two decades, married households in the United States received increasingly more public insurance against labor income risk, whereas the opposite was true for single households. To evaluate the welfare consequences of this trend, I perform a quantitative analysis. As a novel contribution, I expand the standard incomplete markets model à la Aiyagari (1994) to include two groups of households: married and single. The model allows for changes in the marital status of households and accounts for transition dynamics between steady states. I show that the divergent trends in public insurance have a significant detrimental effect on the welfare of both married and single households.

JEL Codes: D52, D60, E21, E62, H31.

The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.