Monetary policy is a primary lever for stimulating consumption and economic growth. However, when central banks hit their limits—whether it’s the lower bound on interest rates, political constraints, or other barriers—fiscal policy must often assume a greater role. But how much fiscal stimulus is needed to offset reduced monetary accommodation? In new research (link), co-authors and I provide an answer: you need fiscal stimulus payments totalling 1.3% of GDP over five years to replicate the economic boost of a one-percentage-point cut in interest rates.
At the onset of the Covid-19 pandemic, monetary policy rates in the United States were at zero per cent. In the euro area, they were negative. Monetary policy was thus significantly constrained. Nevertheless, the pandemic represented a major negative economic shock, necessitating substantial economic support.
With limited room for further monetary easing, fiscal policy had to take the lead. Governments responded with unprecedented stimulus payments (and other fiscal interventions). But this raises a critical question: how much fiscal stimulus is needed to generate the same economic effect as an interest rate cut, had central banks been able to do so?
This question is important for calibrating the appropriate level of economic stimulus and the appropriate monetary-fiscal policy interaction.
The question is at the same time far from easy to answer. Yet, Ulf Nielsson (link), Farzad Saidi (link), Fabian Seyrich (link), Daniel Streitz (link), and I set out to investigate it in new research (link). In this post, I summarise our main findings.
What we do?
Inspired by recent theoretical insights (link, link), our main contribution is to empirically identify a fiscal stimulus payment policy that generates the same increase in private-sector net excess demand as a change in the monetary policy rate.
To achieve this, we first estimate the slopes of aggregate consumption functions. We derive these from microeconomic causal variation—that is, by analysing household-level responses to either interest rate changes or income gains. By combining these two estimated slopes, we can then construct the necessary mapping between fiscal and monetary policy instruments.
In plain language, this means we estimate: (i) how much households adjust their consumption in response to a monetary policy shock; and (ii) how much they change their consumption when they receive a sudden increase in income (similar to a fiscal stimulus transfer). We then aggregate these responses across the economy and map them.
If you think this sounds straightforward, I assure you it is not. Solving this problem requires detailed, high-quality data and careful analysis. Specifically, we need household-level data on consumption, as well as methods to identify how consumers react to both monetary policy and income shocks. Fortunately, in Denmark we have access to precisely such data.
Consumption effects of monetary policy
Our first step is to estimate how much household consumption changes in response to a change in the monetary policy rate.
The Danish setting offers some advantages over other contexts typically examined in the literature. Firstly, the interest rate that borrowers pay on Adjustable-Rate Mortgages (ARMs) is closely tied to the monetary policy rate set by the Danish central bank, as shown in Figure 1. This means that we can study the effects of monetary policy on households by examining changes in the interest payments of ARM mortgage holders.
Figure 1. Yield on short-term mortgage debt and the monetary-policy rate of the Danish central bank (Certificate of Deposit). Source: Figure B.2 here (link).
Secondly, because of Denmark’s fixed exchange rate policy towards the euro, the Danish monetary policy rate closely follows the policy rate set by the European Central Bank. This allows us to treat it as an exogenous source of variation in the interest rates paid by Danish mortgagors.
Thirdly, thanks to the design of the Danish mortgage market, we can compare individuals whose mortgage interest rates reset at different points in time. Adjustable-rate mortgages that reset every three years (so-called F3 mortgages) and every five years (F5 mortgages) are widely used in Denmark (our analysis covers 187,332 F3 borrowers and 421,092 F5 borrowers). Studying these borrowers enables us to identify within-borrower variation in consumption responses.
Moreover, we observe continuous mortgage-rate changes, which is crucial for quantifying the consumption response to interest rate changes. That is, we not only know whether a borrower’s mortgage rate resets but also the exact size of this reset. Finally, we have access to complete income, wealth, and related data for every individual, allowing us to calculate consumption at the individual level.
Equipped with this rich dataset, we estimate how an average borrower adjusts their consumption after experiencing a change in their mortgage rate (driven by a change in the policy rate). The results are presented in Figure 2.
At year t = 0, we find an effect of 3.6%. This means that when a mortgage borrower experiences a one percentage point reduction in their mortgage payment—triggered by a one percentage point cut in the Danish monetary policy rate (following a corresponding cut by the ECB)—they increase their consumption by 3.6%. This provides a robust, micro-founded estimate of the impact of monetary policy on household consumption.
Figure 2. Consumption response to reduced mortgage payments for year t = 0 (the year the mortgage payment is changed due to a monetary policy shock), year t = 1, etc. Source: Table 2 here (link).
The remaining columns in Figure 2 present the dynamic effects on consumption one to four years after the mortgage rate reset. We observe a hump-shaped pattern: the consumption response initially increases further one year later but then falls to well below the contemporaneous response from year 2 onwards, ultimately reaching 1.7% in year 4 (last column).
Consumption effects of income shocks
A fiscal stimulus payment takes the form of a cash payment to households. To estimate the consumption response to such payments, we exploit how households react to unexpected, lump-sum windfall income gains. Specifically, we focus on sudden inheritances arising from the unexpected deaths of close relatives.
We look at both sudden natural (cardiac arrest, stroke, etc.) and non-natural deaths (vehicular accidents, exposure to harmful substances, etc. but exclude suicides and violent assaults, as these might be anticipated). This may sound somewhat morbid to those unfamiliar with such studies, but it illustrates the richness of our data and the unique insights it offers. It is important to note that all data are fully anonymised in the registries, meaning we cannot identify the individuals involved.
We focus on deposit-only inheritances to match fiscal stimulus payments. We then estimate how much household consumption changes when an unexpected cash inheritance is received. The results are presented in Figure 3.
Figure 3. Consumption response to unexpected income shocks due to cash-only inheritances for year t = 0 (the year the inheritance is received), year t = 1, etc. Source: Table 2 here (link).
The columns in Figure 3 show Marginal Propensities to Consume (MPCs), which measure how much consumption changes when income increases by 1 Danish krone, at the year of the inheritance (year t = 0), one year later, and so on, up to four years after the inheritance.
At the year of the inheritance, households consume just over 40% of the windfall. This represents a robust, micro-founded estimate of the effect of an unexpected income shock on household consumption. We find that the MPC falls to 30% one year later, and then continues to decline, reaching 6.5% by year t+4.
Aggregated consumption response to monetary policy
We have now estimated how individual households respond to both a monetary policy shock and an income shock. While this is already impressive, it is not our final objective. Ultimately, we aim to identify the size of a fiscal stimulus payment that would generate the same aggregate consumption response as a change in monetary policy.
To do this, we first use our micro-founded estimates to calculate the aggregate consumption response to a monetary policy change.
To illustrate, suppose a borrower in a given F-loan category has annual consumption of DKK 200,000 in year t−1, i.e. the year before the interest rate change. We then assume that this borrower increases their consumption by DKK 7,200 (= 3.6% × 200,000) in response to a 1 percentage point reduction in the monetary policy rate in year t = 0, where 3.6% is the contemporaneous response shown in Figure 2. In year t+1, their consumption is DKK 8,000 (= 4.0% × 200,000) higher than in the year before the interest rate change, and so on. We then aggregate these consumption responses across all individuals in each F-loan category.
We find that following a monetary policy shock, aggregate consumption rises by DKK 5.6 billion in year t = 0, DKK 6.2 billion in year t+1, and so forth.
The total consumption in Denmark among all individuals aged 18 and over was DKK 952 billion in 2017. This means that—because of a monetary policy shock and the effect this shock has on households’ mortgage payments—aggregate consumption increases by 0.6% (= 5.6/952) in year t = 0, by 0.7% in year t+1, by 0.4% in years t+2 and t+3, and by 0.3% in year t+4.
To compare these effects to the total consumption effects from monetary policy, including—including indirect general equilibrium effects—we compare these to results from Structural VAR analyses. These results are displayed in Figure 4.
Figure 4. The quarterly impulse responses of aggregate real consumption to a 1 p.p. increase in interest rates, derived from SVAR estimates for Denmark. Source: Figure 5 here (link).
We find an aggregate consumption response of around 2–3% over the first year (after four quarters, as shown in the figure). Comparing this to our micro-founded estimate, which showed a 0.6–0.7% effect after one year, suggests that the direct effect of nominal interest rate changes on consumer spending accounts for up to one-third of the total aggregate consumption response to monetary policy.
Demand equivalence
With these results in hand, we are now ready to calculate our ultimate objective.
Recall that we estimated a total consumption effect from monetary policy of DKK 5.6 billion in the first year. This corresponds to a per-person consumption effect of DKK 1,223 (the consumption response of 5.6 billion divided by Denmark’s adult population of approximately 4.6 million).
Next, we ask: what sequence of stimulus payments would be required to generate the same per-person consumption response? From our income/inheritance shock analysis, we found a marginal propensity to consume (MPC) of 41.1% in the first year. This means that a stimulus payment of approximately DKK 2,975 (1,223/0.411) is required to induce a consumption response of DKK 1,223—equivalent to that generated by a monetary policy stimulus.
We perform similar calculations for subsequent years, accounting for delayed consumption responses to stimulus payments made in prior years.
Overall, ignoring discounting, a total stimulus payment of DKK 6,582—equivalent to USD 1,013—paid out over five years per person is needed to replicate the same aggregate consumption trajectory as a 1 percentage point reduction in the monetary policy rate. This is our key finding.
To put this figure into perspective, if such payments were made to the entire Danish population, the aggregate stimulus would amount to approximately DKK 30.15 billion (4,580,547 × 6,582), equivalent to 1.3% of Denmark’s GDP in 2018. The payment in the first year alone would already total 0.6% of GDP.
We, of course, conduct various robustness tests of this calculation, and the results are quite stable.
Income heterogeneity in monetary policy
In the calculations above, we assumed that all households react similarly to monetary policy and income shocks. That is, when estimating the aggregate consumption effect of monetary policy, we applied the average consumption response of 3.6% in the first year uniformly across households. However, some households may respond more strongly than others, while some less so.
To explore the implications of such heterogeneity, we divide mortgage holders into income quintiles and estimate the contemporaneous consumption effect of monetary policy for each group. The results are presented in Figure 5.
Figure 5. Estimates from contemporaneous consumption-response regressions, separately for each disposable-income quintile, alongside 99% confidence intervals. Source: Figure 6 here (link).
We find that lower-income households (those in the first income quintile) respond more strongly to a monetary policy easing: on average, households in the lowest disposable income quintile increase their consumption by over 5%, whereas those in the highest income quintile increase consumption by only around 3%.
However, higher-income households are much more likely to hold mortgages, particularly adjustable-rate mortgages (ARMs), making them more exposed to monetary policy changes through the mortgage channel. Only between 6% and 9% of households in the lowest three income quintiles hold ARMs, compared to 21% in the fourth quintile and 47% in the highest quintile. This means that, overall, monetary policy via the mortgage channel affects high-income households more substantially.
Despite these differences, accounting for income heterogeneity makes very little difference to the total aggregate effect of monetary policy. When incorporating disposable income heterogeneity, aggregate consumption increases by DKK 5.35 billion in the first year, compared to DKK 5.60 billion when we ignored heterogeneity. Since the aggregate consumption effect drives the stimulus calculation, this implies that the demand-equivalent fiscal stimulus path remains largely unchanged.
We also examine other dimensions of heterogeneity than income, including levels of liquid wealth, the share of transfers in total income, age, and geographic location. Across these various factors, the aggregate consumption response remains remarkably robust. From this, we conclude that heterogeneity has minimal impact on estimating the partial-equilibrium effect of monetary policy.
Finally, we assess heterogeneous responses to income shocks (unexpected inheritances) and find relatively flat marginal propensities to consume across both income and liquid wealth distributions.
Our finding that the demand-equivalent stimulus is largely independent of heterogeneous effects is important from a practical policy perspective, as it suggests policymakers need not be overly concerned about such heterogeneity when trying to replicate monetary policy via fiscal stimulus transfers, thereby making their job easier.
Conclusion
Coauthors Ulf Nielsson, Farzad Saidi, Fabian Seyrich, Daniel Streitz, and I have undertaken new research on ‘empirical monetary-fiscal equivalence’ (link).
It represents a substantial piece of work. First, our unique Danish setting allows us to identify a consumption increase of 3.6% in the first year following a one percentage point change in the monetary policy rate, and to track the dynamic impact as well.
Second, we estimate marginal propensities to consume (MPCs) around 40% in the first year following an unexpected income shock. We also estimate the dynamic impact.
However, our contribution goes further: by combining these estimates, we provide the first micro-econometric based empirical evidence on how much fiscal stimulus is needed to replicate the household consumption effect of a one percentage point change in the monetary policy rate. Our main specification finds that stimulus payments totalling 1.3% of GDP over five years are required to mimic the consumption effect of such a monetary policy change.
For comparison, the initial round of coronavirus stimulus checks in the US in 2020 amounted to approximately $388 billion, or 1.9% of US GDP that year. We hope our results will inspire policymakers considering the relationship between fiscal and monetary policy as tools for economic stabilisation.
