leverage will not pass the regulatory requirements when the risk of their credit portfolio
is assessed. If risk-assessment is perfect, the first-best allocation can be achieved in the
MC economy.
At a more detailed level, we start with a two-period, two-sector economy with risk-
neutral agents as in Gersbach and Rochet (2012, 2017), extended to heterogeneous banks
and with asymmetric information of households about individual bank characteristics.
Households and bankers are endowed with a capital good, which they supply to firms in
two sectors in order to produce a consumption good. In the first sector of the economy,
firms have direct access to the capital good through issuing bonds to households. In the
second sector, firms can only obtain capital through bank loans. Banks partly finance
their loans through their own endowment with capital goods, i.e., through equity, and
partly either by household funding (in the LF economy) or by money creation (in the
MC economy). Banks are subject to moral hazard in the spirit of Holmstr¨om and Tirole
(1997). If they monitor loans diligently, their investments are more likely to succeed. If
they shirk monitoring, they enjoy private benefits. Banks are heterogeneous regarding
the benefits from shirking or, equivalently, regarding their efficiency in monitoring.
In the LF economy, the amount of funding households are willing to provide to banks
is limited, since banks’ monitoring incentives decrease proportionally to external financing
and thus to the scale of the bank. With heterogeneous banks and asymmetric information
between households and banks, households limit funds to banks, such that even the
bank with the greatest potential benefits from shirking still monitors. As a consequence,
aggregate external financing of banks, and thus aggregate lending by banks, is low. It is,
of course, lower than in a first-best world without any frictions and also lower than in a
second-best world where the characteristics of a bank are known to households. It turns
out that it is inefficiently low since in the MC economy and with the same informational
frictions, aggregate bank lending will be higher and closer to the second-best outcome.
In the MC economy, banks do not require household funding to initiate lending. Any
loan they hand out simultaneously creates a deposit for the borrower. Firms use the
deposits obtained through loans to buy the capital good from households, which are
credited with deposits at their bank in return. As firms and households are likely to
hold accounts at different banks, the ensuing interbank transactions have to be settled
by reserves. Only banks can borrow such reserves from the CB.
As long as the profits on new loans exceed the bank’s funding costs, increasing money
creation, and thus leverage, is always profitable for an individual bank in the MC econ-
omy, since it increases the bank’s expected return on equity. High leverage, however,
implies low monitoring incentives. Hence, the government acting as a bank regulator
imposes a leverage constraint.
3
By setting this leverage constraint, the regulator aims to
3
Note that our rationale for a maximum leverage ratio, that is, forcing banks to keep enough skin
in the game to guarantee a certain level of aggregate monitoring, is different from the systemic-risk
3
Electronic copy available at: https://ssrn.com/abstract=4095335
