Research

Job Market Paper

Interest Rates vs Unemployment Risk: What Drives Automobile Purchases?

Durable expenditure over the business cycle is thought to be highly dependent on interest rate changes, but recent heterogeneous agent models have revealed unemployment risk as a key determinant of spending by less-wealthy households. Using U.S. macroeconomic data together with household surveys from the Michigan Survey of Consumers and the New York Fed Survey of Consumer Expectations, I provide evidence that subjective household unemployment expectations are a stronger predictor of durable expenditure than interest rates. In recessions, when households expect unemployment risk to rise, they postpone durable purchases until the outlook stabilizes. Motivated by these findings, I build a heterogeneous agent model with both durable and non-durable goods. Using elicited changes in subjective expectations, I decompose the response of durable expenditure into components driven by unemployment risk and interest rates. A variance decomposition attributes roughly 80% of fluctuations in durable expenditure to unemployment risk - of which 50% reflects increased uncertainty (second moment) and 30% the expected decline in average income (first moment) - and 20% to interest rates. These results highlight the dominant role of unemployment risk in shaping household durable expenditure and underscore the importance of incorporating subjective expectations into heterogeneous agent models.

Working Papers

Perceived Unemployment Risks Over
Business Cycles (Bank of Canada WP)

with Will Du (JHU), Xincheng Qiu (Peking University), and Tao Wang (Bank of Canada)

We backcast subjective expectations on job finding and separation in the Survey of Consumer Expectations to 1978, and use real-time machine learning forecasting to proxy their objective counterparts. We document stickiness in job finding and separation expectations in reflecting changes in real-time job finding and separation risks and their substantial heterogeneity across observable and unobservable dimensions. Calibrating these facts into a heterogeneous-agent consumption-saving model reveals that belief stickiness attenuates the precautionary saving channel. As a result, workers under-insure during recessions, leading to a more sluggish recovery afterwards. The combination of high-risk exposure and under-insurance due to belief stickiness operates as a novel amplification mechanism over the business cycle.

Seeing the Economy with Colored Glasses: Partisanship in Macro and
(not in) Micro Expectations (link)

with Kyung Woong Koh (JHU) and Tao Wang (Bank of Canada)

Political views affect households' macroeconomic expectations, but personal economic circumstances and self-interest remain the dominant factors shaping their beliefs. Using an expanded dataset covering 11 U.S. Presidential elections from 1980 to 2020, we show that households’ personal finance expectations exhibit significantly less partisan bias than their macroeconomic expectations, as households are more directly informed about their own situations. We show that households "cheerlead" for policies to be beneficial to the broader economy often not because such policies are enacted by their favored winning party, but because they expect to personally gain from them. We develop and empirically estimate a factor model of belief formation that integrates mechanisms such as partisan biases, political sentiment, and differences in belief extrapolation. Our analysis quantifies the time-varying importance of partisanship and microeconomic disparity in driving polarized views of the macroeconomy.

Beliefs, Precautionary Savings and
Home Ownership (ECB WP)

with Johannes Fleck (Fed Board)

This paper shows that individual beliefs on the effectiveness of formal and informal sources of risk sharing determine financial precautionary behavior. We present empirical evidence demonstrating that higher trust in public insurance systems reduces net liquid wealth while higher trust in communal insurance increases it. This dichotomy is consistent with theories on access to private risk sharing networks. Moreover, we find that both types of trust associate positively with the probability to take on financial risk for the purpose of becoming a homeowner and the related loan-to-value ratio. Our findings are robust across a widerange of econometric controls and specifications.

Work in Progress

Automobile purchases and unemployment risk - Evidence from Swedish Administrative Data
with Jesper Böjeryd and Roine Vestman (Stockholm University)

Can we quantify the impact of rising perceived unemployment risk on automobile purchases? From administrative data, we extract the exact month and value of individual household automobile purchases for the entire Swedish population from 2000 to 2019. To proxy subjective unemployment risk, we exploit cross-sectoral variation in firms’ expectations of future employment from the Swedish Tendency Survey. Assuming workers internalize sectoral employment outlooks when forming expectations about their own job security, we assign sector-specific risk measures to individuals based on their occupation. This allows us to estimate the causal effect of changes in perceived unemployment risk on car purchases over the following year. We document impulse-response-type effects of unemployment risk on both the extensive margin (whether to buy a car) and the intensive margin (how much to spend). Our results provide the first month-level quantitative evidence of behavioral response to unemployment risk and offer micro-level moments for disciplining heterogeneous agent models with durable goods.

Fiscal Consolidation with Informal Labor Markets
with Yongquan Cao and Carlos Goncalves (IMF)

High public debt levels have prompted many countries to pursue fiscal consolidation. This paper studies such efforts in the context of developing economies with large informal labor markets that are difficult to tax. We extend a standard Heterogeneous Agent New Keynesian (HANK) model by introducing two labor market sectors — formal and informal — with exogenous worker transitions. Consistent with empirical evidence, formal sector jobs offer higher wages and lower unemployment risk but are subject to income taxation, while informal sector jobs are riskier but untaxed. We use this framework to analyze the aggregate and distributional effects of fiscal consolidation. Our results highlight how the presence of informality shapes the transmission of fiscal policy and creates trade-offs between efficiency, equity, and fiscal capacity.

Equity Premium Puzzle, Expectations and Covid-19: Evidence from Germany

What differentiates households who invest in financial assets from those who do not? Using data from the Bundesbank Online Pilot Survey on Consumer Expectations in Germany during the COVID-19 pandemic, I directly elicit households’ self-reported reasons for their investment decisions. I find that the main barriers to participation in financial markets are a lack of information and a lack of interest, while time constraints and risk concerns primarily affect households’ ability to re-balance existing portfolios. Among buyers, investment is typically driven either by expectations of rising asset prices or by adherence to a pre-existing savings plan. For the former group, as well as for first-time investors, having additional time is a key enabling factor. Finally, I document that households with higher expectations of property price growth or inflation are less likely to invest in financial assets, suggesting perceived relative returns influence portfolio allocation.

Do Homeowners Save more than Renters? – Evidence from the German Wealth Survey
with Julia Le Blanc (European Commission) and Tobias Schmidt (Deutsche Bundesbank)

We analyze how property ownership affects household saving behavior using three waves of a large survey on household finances in Germany that allow us to track both financial asset accumulation and mortgage repayments as components of active saving. We find that homeowners with mortgages save significantly more than renters, if mortgage payments are included in active saving flows. Excluding mortgage payments, the difference in financial saving flows between renters and mortgaged owners is small and statistically insignificant. This suggests that households do not substitute between mortgage repayments and financial savings but instead treat repayments as additional saving. Moreover, homeowners who have fully repaid their mortgages display intermediate saving behaviors — saving more than renters but less than mortgaged owners — suggesting that repaying a mortgage induces “learning”. These findings have implications for the aggregate response of consumption and saving to macroeconomic shocks, particularly given that homeowners exhibit systematically different marginal propensities to consume, even when controlling for liquidity constraints (hand-to-mouth status).

Page updated

Google Sites

Report abuse