We study foreclosure discounts for all forced Norwegian house sales between December 2001 and February 2018. Micro-data from two Norwegian banks are further analyzed to shed light upon biased valuation methods. We estimate foreclosure discounts using an advanced statistical valuation model (AVM), repeated sales, and a real estate agent’s value estimate. We expand on earlier research by exploring how homeowner participation, centrality and the valuation method (human vs. automated valuation) adopted at origination are interlinked and contribute to the explanation of foreclosure discounts. Homeowners in rural areas are less likely to participate in foreclosure sales, and banks granting mortgages are less likely to employ AVM valuation. Our findings enrich research on biased valuations and link the bias in valuation methods with bias in banks’ valuation methods. Banks granting mortgages with a high loan-to-value ratio should exercise caution in rural areas while deviating from AVM valuation and calibrating their lending models considering risk differences.