From Evidence to Policy: What the Model Can and Cannot Say

The empirical evidence shows that young and small firms can be exposed through both collateral and local credit supply, but it does not observe every balance-sheet interaction at once. This chapter examines the quantitative model built to connect those pieces: what the model isolates, how it produces a persistent young-firm employment decline from a housing shock, and what its policy counterfactuals do and do not establish. The useful judgment is conditional rather than absolute: the model can compare channels under shared assumptions, not certify a particular intervention as costless or universally effective.

A causal laboratory for two balance sheets

The model keeps the minimum structure needed to make the two financing routes interact. Households choose consumption, labor, and owner-occupied and rental housing. Entrepreneurs operate firms that use labor and capital, can borrow, and face a constraint tied to the capital they own and the value of aggregate housing. Banks take deposits and hold mortgages, business loans, and rental housing. A housing producer supplies the housing stock. Entry and exit are included so that the model can speak to firm creation and liquidation rather than only to the employment of a fixed set of firms.

The firm-side constraint is the direct collateral route. A firm can borrow against productive capital and a fraction of housing collateral. When the house price falls, the housing component of its borrowing capacity falls. The firm may have to reduce investment or labor demand, and a potential entrant may begin with less capital. The constraint is especially consequential for a young firm whose internal assets are small or whose borrowing limit is already close to binding.

The bank-side constraint is the indirect credit-supply route. Banks own assets whose values and returns are affected by housing. Their net worth is the value of assets minus deposits. An internal enforcement condition limits how far a bank can leverage its net worth. When housing prices reduce bank net worth, the bank’s permissible leverage falls and the lending spread rises. The firm then faces a higher borrowing cost even if its own housing collateral has not changed by the same amount.

The two routes meet in the firm’s operating decision. A lower collateral limit reduces the quantity of funds available; a higher lending rate raises the cost of funds and can change the desired scale of labor and capital. A firm may therefore cut hiring because it cannot borrow enough, because borrowing has become too expensive, or because both conditions hold. The model does not need to assign every observed job loss to a single route. It needs to keep the routes separate so that each can be switched off in a counterfactual.

The shock is deliberately placed outside production and finance. It is modeled as a temporary, unanticipated decline in household preferences for housing in general and owner-occupied housing in particular, followed by gradual recovery. This choice matters. There is no direct shock to total factor productivity, the firm’s collateral parameter, or the bank’s enforcement parameter in the main experiment. House prices fall because households want less housing; the production sector and financial sector then transmit and amplify that movement through the two balance sheets.

That setup is not a claim that the Great Recession was caused only by a household preference change.