The macroprudential toolkit: effectiveness and interactions

Stephen Millard, Margarita Rubio and Alexandra Varadi

The 2008 global financial crisis showed the need for effective macroprudential policy. But what tools should macroprudential policy makers use and how effective are they? We examined these questions in in a recent staff working paper. We introduced different macroprudential tools into a dynamic stochastic general equilibrium (DSGE) model of the UK economy and compared their impact on the economy and household welfare, as well as their interaction with each other and with monetary policy. We found that capital requirements reduce the effects of financial shocks. Instead, a limit on how much of borrowers’ income is spent on mortgage interest payments reduces the volatility of lending, output and inflation resulting from housing market shocks.

What we do

We developed a DSGE model with capital requirements for banks and two housing tools applied on mortgage borrowing:  a loan to value (LTV) constraint and a ‘stressed’ debt service ratio (DSR) limit. The stressed DSR limit tests whether borrowers have sufficient income to afford their mortgage payments (ie, remain within the DSR limit) should interest rates rise. With the exception of the interest rate effect, the stressed DSR limit in our model is similar to a loan to income limit. The LTV and DSR housing tools in our model affect borrowing households, who take mortgage loans from banks. In contrast, saving households hold bank deposits, own housing, and have no debt. The model also features firms, which borrow from banks to finance their working capital needs and face costs of adjusting prices. Banks face macroprudential capital requirements and complying with these causes them to incur costs, which rise exponentially the nearer the bank is to the limit. Following Gertler and Karadi (2011), we introduced information frictions between bankers and depositors, which generates a constraint on banks’ leverage, and in turn, a spread between deposit and lending rates. 

To examine the interactions between the various macroprudential policies and monetary policy, as well as their impact on economic performance, we compared four versions of the model. Our baseline model included an LTV limit. Model 2 added capital requirements. Model 3 was a version of the baseline with a DSR limit rather than an LTV limit. And Model 4 contained a DSR limit (rather than an LTV limit) and capital requirements.

Adding capital requirements to an LTV limit alleviates the impact of financial shocks on the lending spread

Chart 1 shows the response of the spread between the lending and the deposit rate in the baseline model, with and without capital requirements switched on. The chart shows that the introduction of capital requirements dampens the response of the lending spread to a negative productivity shock, a housing demand shock, and a shock to non-performing loans (proxying a more general financial shock). For instance, the lending spread rises by 1,844 basis points in the baseline model following the financial shock, compared to only 14 basis points once capital requirements are imposed. This implies that capital requirements are able to insulate the real economy from the effects of a financial shock, since the lending spread is the main channel through which these shocks are propagated to firms and households.

Chart 1: Behaviour of bank lending spread

An LTV limit is less able than a DSR limit to constrain borrowing in booms, unless augmented by capital requirements

Chart 2 shows the implied responses of the DSR and LTV ratios following a housing demand shock that increases house prices by 3%. For the models with the LTV limit switched on (top panels), we calculate the prevailing DSR in the economy (blue line). For the versions of the model where the DSR limit is switched on (bottom panels), we calculate the prevailing LTV ratio in the economy (red line). In the top panel, where household finance is constrained by an LTV limit, the DSR jumps significantly upon impact. The house price appreciation supports more borrowing for the same collateral constraint, resulting in higher spending and a subsequent monetary policy tightening to bring inflation back to target. The monetary policy tightening increases the costs of servicing the debt, and has a direct effect on the DSR in the economy. The DSR is less responsive to the housing demand shock when the macroprudential LTV limit is augmented by capital requirements (top-right panel of Chart 2). This occurs because lending and interest rates respond less to the shock when capital requirements are imposed on banks. 

The bottom two panels show that the implied LTV ratio when macroprudential policy operates through a DSR limit (with or without capital requirements) remains relatively unchanged following the house price shock. This occurs because the DSR limit breaks the link between house prices and mortgage borrowing. As a result, a house price appreciation does not translate into a loosening of household credit constraints, as these are, instead, linked to labour incomes.

Hence, to limit credit booms, policymakers would have to augment an LTV limit with either tight capital requirements or with a DSR limit. In contrast, capital requirements add little to the effect of a DSR limit on debt in booms.

Chart 2: Performance of housing tools following a housing demand shock (3% rise in prices)

Monetary policy responds less in booms when DSR limits are switched on

Chart 3 and Chart 4 show the base rate response following positive technology and housing demand shocks in the four models discussed above. In both experiments, monetary policy responds less to shocks when a DSR limit is imposed. 

A technology shock (Chart 3) leads to higher output and lower prices, and in turn to looser monetary policy. With a DSR limit in place, this directly loosens household borrowing constraints by lowering the costs of servicing debt. As a result, household borrowing increases, leading to stronger economic activity and a less severe impact on inflation that requires a less aggressive response from the monetary policy maker. This suggests that, when the economy experiences a technology shock, a DSR limit may support the objectives of the monetary policy maker. In contrast, if an LTV limit is in place, monetary policy has to be more active in order to bring inflation back to target.

Chart 3: Monetary policy rate following a technology shock (0.5% rise in GDP)

Following a housing demand shock (Chart 4), monetary policy tightens when an LTV limit is imposed, due to the positive feedback between borrowing, collateral constraints and higher spending. However, the base rate responds less when capital requirements complement the LTV ratio (green versus black lines). This occurs because capital requirements dampen the effect of the house price shock on lending, which decreases the effect of the shock on GDP and inflation. Hence, macroprudential policy acting through capital requirements contributes to price stability in the face of a housing demand shock.

Nevertheless, monetary policy remains unchanged when lending to households is constrained by a DSR limit (blue and magenta lines). Since a DSR limit breaks the link between borrowing and housing wealth, a shock to house prices does not influence how much households can borrow. 

Chart 4: Monetary policy rate following a housing demand shock (3% rise in prices)

Relative to an LTV tool, a DSR limit reduces macroeconomic volatility and improves household welfare following a housing demand shock

We show that a DSR limit is the most effective tool for reducing the impact of housing market shocks, by comparing the volatility of key macroeconomic variables in our four models. Table A shows that, relative to a model with LTV tools with or without capital requirements, a DSR limit leads to a significant decrease in the volatility of lending, consumption and inflation since it insulates the real economy from housing market shocks. To evaluate the impact of our macroprudential tools on household welfare, we derive a welfare-based loss function using a weighted average of the utility functions of savers and borrowers in our model. We then assess which tools minimise the welfare loss. Table A illustrates that a DSR limit improves welfare by more than the LTV tool alone, or with capital requirements in place. In addition, adding capital requirements to an existing DSR limit makes little difference in terms of the welfare loss (final row). 

Table A: Volatility and welfare loss following a housing demand shock (3% rise in prices)


We found that DSR limits are more effective than LTV tools and capital requirements at reducing the volatility of economic variables following a housing demand shock. In addition, there is a direct link between DSR limits and monetary policy via the impact of base rate changes on the cost of servicing mortgage debt. As a result, DSR tools can help the monetary policy maker during a boom.

In contrast, our work suggests that capital requirements are better suited at addressing real-economy risks from negative financial shocks as they reduce the response of the lending spread to such shocks. Also, capital requirements can provide an effective complement to an LTV limit in constraining household indebtedness in booms and keeping DSRs under control.

Stephen Millard works in the of the Bank’s Structural Economics Division, Margarita Rubio works at Nottingham University and Alexandra Varadi works in the Bank’s Macroprudential Risks Division.

If you want to get in touch, please email us at or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.