Back-of-the-envelope low-quality musings: #takes

Longform stuff I've put time and thought into is at #effortposts.

To hear from academics and policymakers, #interviews.

Compiled scribbles on what I read can be found in #notes.

Topic-related tags include #economics, #coronavirus and #miscellaneous.

# The Next Recession: Where Do We Go From Here?

05 Sep 2020

Disclaimer: this has been tagged as an effortpost due to the time and work put in. However, do take the monetary and infrastructure prescriptions with a pinch of salt. There’s a reason this is still an open debate among people far more qualified than myself.

TL;DR (because I don’t know who (if anyone) reads my blog and I certainly don’t expect most people to go through nearly 6,000 words of optimal stabilisation theory): Our business cycle stabilisation policy should aim to maintain low and stable inflation as well as minimise the output gap. The pre-2008 consensus was that the use of the flexible inflation targeting regime within monetary policy was enough to do the job. But the lowering natural rate of interest has meant that we are more likely to hit the effective lower bound, while central bankers have proven that distinguishing between demand and supply-side shocks is difficult to do in real time. These constraints on our monetary policy tools mean we need to update our toolkit. In times of significant downturns, we should temporarily target the nominal GDP level for the foreseeable future. This is inherently countercyclical, dampens supply-side shocks and is effective at keeping us from the effective lower bound. We should also make greater use of automatic fiscal stabilisers, and provide universal cash transfers when the Sahm Rule indicates we are in recession. Finally, all of this should be buttressed by providing savings accounts at central banks to the entire population, allowing the monetary and fiscal mechanisms to be more effective.)

Alternatively: Skip past the first three sections to get to the juicy stuff on policy prescriptions! But skip at your own peril - there’s a reason the history and context is there.

2020 marks the second time in under 15 years that we have had an out-of-sample economic crisis. And though the response this time was improved from 2008, which was in turn better than in 1929, it hasn’t been perfect - the recent update of the Federal Reserve mandate and the release of the Bank of England’s research agenda are a testament to that. It is particularly frustrating to see some of the same issues crop up again and again, due to political considerations and economic constraints. It therefore is worth retracing our steps - to see why stabilisation policy is the way it is, where it has fallen short of our goals and how it can be improved.

What is the aim of stabilisation policy?

The two macroeconomic variables we are usually concerned about in stabilising business cycles is inflation and output. When inflation is high, changes in the relative prices of goods and services is obscured by the general rise in the price level. This makes resource allocation based on the information provided by prices less efficient. High inflation also means that there are menu costs, where businesses need to change their prices more frequently. Variable inflation means that investors buying assets must be compensated for with an inflation risk premium, since the returns on their assets are in nominal terms that could be eaten away by inflation. This makes asset pricing inefficient. And unanticipated inflation arbitrarily redistributes wealth and income. For example, there is a redistribution from lenders to borrowers, since the money borrowers need to pay back is worth less due to inflation. As such, we want low and stable inflation.

Economic output in the short run is determined by the intersection of the aggregate supply and aggregate demand in the economy. But in the long run, it is determined by the level of potential output the economy can produce when it is fully utilising its resources. When the economy is operating at its potential output level, the unemployment level is defined as the natural rate of unemployment $u*$ and the real interest rate is the Wicksellian/natural rate of interest $r*$. The natural interest rate simply represents the marginal return on capital - for fairly obvious reasons, if the returns on investing in something were higher than the costs of borrowing, people would borrow to invest until the two equalised. Because there are various frictions and rigidities within markets, the actual level of output in the short run does not necessarily equal the long-run potential output. This creates an output gap, with inflation occurring if there is a positive gap and unemployment if there is a negative gap. As such, we want to minimise the output gap.

Inflation is always and everywhere a monetary phenomenon - that is, it simply represents too many dollars chasing after too few goods and services. Consequently, we can control it via monetary policy.

The output gap is influenced by the amount of aggregate demand in the economy. AD, which represents people’s choice of whether or not to spend, depends on the real interest rate. And so the traditional approach of minimising the output gap is by adjusting the real interest rate relative to where we think the natural rate is. If we are faced with an negative AD shock where spending in the economy drops, we could lower the real rate below the natural rate. This makes the cost of borrowing lower than the returns on investment, boosting AD back up. The Fisher equation $R_t = i_t - \pi_t$ tells us that the real interest rate is the nominal interest rate minus inflation. Because prices and inflation adjusts slowly over time, central banks can adjust the nominal interest rate to have an effect on the real interest rate in the short run.

It seems straightforward to think of unemployment as a useful proxy for the output gap, especially since it is easier to measure that in real time than come up with some abstract notion of what the theoretical potential output of an economy is. The Phillips Curve says that there is a negative relationship between unemployment and inflation in the short run. The combination of these two facts creates the divine coincidence1 that targeting inflation allows us to target both goals at the same time.

The consequence of all of this is that stabilisation policy since the 1990s in developed economies is the flexible inflation targeting regime. That means central banks aim for some long-run inflation rate $\pi*$. When the two goals come into conflict, central bankers take a balance approach towards controlling inflation and the output gap. Much like sailors finding their way home using celestial bodies, “policymakers should navigate by the stars”2 of $u*$, $r*$ and $\pi*$. The consensus was also that fiscal policy was not necessary for stabilisation purposes - not only was monetary policy adequate, fiscal policy was subject to all sorts of time lags and political considerations3

How is monetary policy constrained?

The foundations of this macroeconomic consensus was shaken to its core in 2008. The premise of this regime is the Phillips Curve, and unfortunately, she is a fickle mistress. Firstly, consider what happens when there is an aggregate supply shock - for example, the price of oil skyrocketing. This pushes inflation upwards but output downwards - a central bank faced with an AS shock is forced to either further reduce output or accept a deviation from the inflation target. Secondly, the sensitivity of inflation to the output gap has reduce over time4. That means the tradeoff when faced with an AS shock has changed too.

Because central bankers are human and fallible, it is difficult for them to differentiate between AS and AD shocks in real time and it is difficult to recognise when these relationships change. For example, the oil shock in late 2008 is one of the reasons the Federal Reserve did not loosen monetary policy and was deterred from mitigating the collapse of AD. In 2011, the Bank of England was two votes away from raising interest rates due to oil shock - that would have been a mistake, as evidenced by the collapse in output when the European Central Bank did that. So flexible inflation targeting is itself quite a difficult business.

A second impediment has been the decline of the natural interest rate $r*$. Slower productivity growth, a rising demand for safe assets and a secular stagnation have all reduced the demand for savings, while an aging population and a global savings glut has increased the supply. Consequently, the natural rate has fallen and persists at a low point of around 1%5. We know from above that if the real interest rate is above the natural rate, monetary policy is contractionary. The decline of the natural rate means there is a cap on the real interest rate during normal times. This gives less room to lower real and nominal rates during recessions, as negative interest rates will lead people to take their money out as cash, which pays 0% interest. The costs of holding cash means there is an effective lower bound, even if it isn’t a zero lower bound. Indeed, short-run rates could be at 0 up to 40% of the time, significantly constraining the conventional instrument of monetary policy6.

A third issue arises due to the diminishing returns on lowering interest rates. When prices change, there are substitution and income effects - for example, if the price of apples falls, they become cheaper relative to other goods and so you may substitute from those goods to buying apples. But your real disposable income also rises, and a richer person may prefer to have more apples. So there are two ways in which a price change affects demand - and crucially, they don’t always work in the same direction. Consider wages as the price of leisure. The substitution effect as wages rise is that the cost of leisure increases, meaning people work more. The income effect as wages rise is that richer people may want to spend more time on leisure, reducing their desire to work. This is why you cannot keep increasing wages and expect people to work more hours.

We can apply this by considering interest rates as the price of consumption in this period, instead of saving it for consumption in the next period. As interest rates lower, the substitution effect is that consumption is cheaper and occurs more. This effect is likely to be stronger for credit-constrained consumers with pent-up demand. It is probably close to 0 now, since borrowing costs are basically not a constraint for businesses8. The income effect of lower interest rates is that maintaining the same level of retirement income requires saving more now. This is exacerbated by increasing lifespans and rising healthcare costs. If the effects on savings swamp those on consumption, lowering interest rates may not generate any more demand right now. (Of course, the transitory nature of monetary policy effects and the wealth effect also matters, but it should made us think more about how effective this is.)

If this conventional policy tool doesn’t work, how about using the unconventional tools we’ve started trying out since 2008? Well, it turns out these aren’t great either. One option is to have negative interest rates. These face the limits of lowering interest rates as mentioned above, as well as the effective lower bound. And since customers can take some money out as cash, this means that retail banks will struggle to pass on these negative interest rates without seriously cutting away at their profits. Important financial intermediaries like money market mutual funds, which promise to ensure that they never lose money, will face difficulties too. Some have suggested abolishing cash or trying to implement a negative interest rate on cash too. But cash exists for a reason - this is likely to create a lot of resistance, encouraging significant resources to be devoted to the wasteful task of avoidance via prepayments or using other currencies. Coupled with the ethical and political issues of the government having a record of all transactions, this is a practical nightmare.

Another more popular option is quantitative easing - this is where central banks engage in large-scale asset purchases. By buying up safer bonds, investors will take the money and buy other riskier assets. This increases their price, creating a positive wealth effect that may encourage more spending. Unfortunately, this also massively expands central bank balance sheets, exposing them to market changes and causing distortions in financial markets. Unsurprisingly, there is a “great deal of uncertainty about the magnitude of these effects and their impact on the overall economy”9.

The failure of QE has led some voices to call for QE for the people, or helicopter money. Popularised by Bernanke’s Japan paper10, this is basically just the central bank giving money to individuals. This isn’t a bad idea, and there are good arguments made by made by Mark Blyth, Eric Lonergan and Simon Wren-Lewis for why it’s useful. But fundamentally, this represents a reasonably big change in the legal mandate and economic role of central banks, as it veers into the territory of fiscal policy.

How is fiscal policy constrained?

So let’s look at fiscal stimulus. Time delay is a significant problem this regard - one of the key principles of fiscal stimulus is that it should be timely11. When fiscal stimulus arrives in a tardy fashion, it means that a greater amount is required as the country sees its economic output fall even further. There may even be hysteresis effects, where a reduction in AD ends up leading to a reduction in the potential output of an economy - for example, people who are unemployed for a long time may lose their skills or never return to the workforce. As such, time is of the essence.

The need for legislative approval provides one source of tardiness. In thhe best case, there is a spirited debate and negotiation over the form of stimulus, driven by ideological differences over the extent to which government deficits are a problem, electoral incentives to engage in pork-barrel politics and disparate personal judgements on whether fiscal stimulus should come from cuts in taxes or increases in spending. This alone can cause a chilling effect - Larry Summers famously ignored Christina Romer’s policy memo asking for a $1.8 trillion stimulus during 2008, because he thought that no one in Congress would ever agree to it. Instead, the policy memo he ended up providing to President Obama included options that didn’t come close to the t-word of trillion12. But in its worst form, it looks like the Tories choosing the politically-expedient but economically-dubious option of engaging in austerity in the UK during the 2010s, or the Republicans deciding not to come up with a compromise but instead letting the Federal Pandemic Unemployment Compensation run out in July. Fiscal stimulus is incredibly vulnerable to these sorts of political machinations. But even once a fiscal stimulus package has been agreed to, there is a delay in getting the money from the legal document into the economy. There are many types of fiscal stimulus under the status quo: tax relief and incentives, spending on welfare programs as well as spending on various projects relating to healthcare, infrastructure, education and so on. Tax cuts are often only useful for people with jobs and enough disposable income, rendering them difficult to target effectively at the most vulnerable. The conditionality of welfare spending makes delays inevitable - the FPUC has made abundantly clear that existing systems are hopelessly inadequate at dealing with the scale of claims, resulting in a significant proportion of people having to wait months to get the fiscal support13. And for infrastructure projects, the fact that we ought not be building random things for the heck of it means that the process of determining which projects are most useful is itself a cause of tardiness, not to mention the inevitable logistical delays in finding a contractor and getting to work. So our existing fiscal stimulus is a pretty lethargic affair. Finally, there is an open debate on the effectiveness of fiscal stimulus itself. One reason is the concept of Ricardian equivalence as formalised by Robert Barro14 - that is, any fiscal stimulus needs to be paid for by taxation or borrowing. And since borrowing now must be funded by taxes later, a forward-looking consumer should realise that they will need to pay higher taxes later. As such, any fiscal stimulus will be offset by the reduction in spending now, as households save up to pay for that tax cut. More generally, the objection of crowding out suggests that government borrowing increases interest rates, making private investment more expensive. How can we do better? I don’t want to litigate the details of whether we need fiscal stimulus, but I will note several things. 1. Even if unconventional monetary tools work at the zero lower bound, we have much less certainty about how well they work and their consequences. That provides a role for fiscal stimulus to complement any monetary response. 2. Even in the old consensus view, where countercyclical monetary policy dominated countercyclical fiscal policy, there was still a role for automatic countercyclical stabilisers15. 3. The objections justified by Ricardian equivalence rely on unrealistic beliefs about households that are entirely forward-looking, rational, infinitely lived, altruistic and credit-unconstrained. 4. The dangers of crowding out are less likely in a world with persistently low interest rates, monetary expansion and a significant amount of slack in the economy. In line with the new view16, I will treat fiscal stimulus as a meaningful part of the response. Not only is this confirmed by various empirical studies171819, but one of my favourite economists, Emi Nakamura, has written a particularly illustrative paper on this with co-author Jon Steinsson. They take advantage of interstate variation in military spending in the US to show that the data is more consistent with New Keynesian models. Those models imply that there are non-trivial fiscal multipliers, conditional upon being in a liquidity trap where monetary policy is expansionary20. These are pretty reasonable conditions when we are faced with a significant recession. Having established the need for both monetary and fiscal tools, we could adjust the specifics of various spending programs or credit/liquidity facilities. But instead of getting stuck in the weeds, I want to look at more radical changes to our stabilisation options. Where does monetary policy go from here? Dr. Ben Bernanke has argued that the incremental way in which current policy options are used is useful, but unlikely to be enough to stop a massive crisis where we are persistently at the effective lower bound, due to the range of criticisms leveled above. So that means we need to look at changing something bigger - altering the target which the central bank aims for. One way to avoid the lower bound is by increasing the inflation target - from the Fisher equation, we know that this should push up nominal interest rates and give us more room. However, this forces society to bear the welfare losses of inflation during all times instead of just around a crisis, as well as forcing the central bank to re-anchor inflation expectations at a higher level. This is not an easy thing to do, especially since inflation is politically unpopular. But perhaps most importantly, any policy response should be calibrated to the severity of the crisis and the amount of time spent at the lower bound - the fact that this varies from recession to recession makes a one-off change in the inflation target suboptimal. Another suggested option is to target the price level - that is, a central bank would target a path for the price level instead of targeting the change in the price level every year. Under 2% inflation targeting, a central bank that faces 1% inflation this year still targets 2% inflation the next year. But with price level targeting, they would need to correct for this in the future, such as by aiming for 3% inflation in the following year, in order to stay on that path. This means they cannot treat bygones as bygones. This can reduce price level uncertainty over a long time horizon - however, it means that there is more variability over the inflation rate within that period of time. It can help deal with AD shocks - the expectation that money will be loose in the future and inflation will be allowed to overshoot the target means that real interest rates in the future will be lower, mitigating the current decline in output and inflation. However, the fact that central banks cannot look through a temporary inflation change means that when faced with an AS shock, the central bank is forced to reverse that effect and engender a contraction in output. The limitations of these two approaches encouraged former Chairman Bernanke to suggest that central banks should commit to temporary price level targeting when the lower bound is reached. To be clear, this means that a central bank when face with the lower bound would commit to a price level target for the next few years, without changing its normal target. This has the same effect on people’s expectations of future inflation as a simple price level target. But is expectations management effective in practice? As it turns out, our existing use of forward guidance has been effective21. Forward guidance comes in two flavours - one is where the central bank forecasts the future like the Oracle of Delphi, and in doing so gestures at what it might do. The other has the central bank committing itself to future action, like Odysseus tying his hands to the mast of his ship. It is the Odyssean rather than Delphic forward guidance that has been most effective22, and a TPLT provides that sort of commitment to allowing excess inflation in the next few years. In fact, it is even better than the existing methods of forward guidance, where a central bank promises to allow inflation to overshoot the target for a few years. Consider what happens in a few years time when inflation is above target - since all the gains from people spending will have been accrued, a central bank has no incentive not to renege and control inflation. Realising this, rational households and firms will not believe the central bank’s promise is credible, undermining its effect in the first place. But a TPLT doesn’t give the central bank any choice in this matter, and that commitment allows it to overcome this time inconsistency problem and make its promises credible. The fact that a TPLT focuses on lowering the generally higher long-run interest rates by higher inflations expectations is helpful, given the limited effectiveness of further reducing short-run real interest rates. And compared to the existing options of lowering long-run interest rates by QE, we know that setting the rate of expected inflation is just as good a stabilisation tool23. In fact, it may be more effective - Paul Krugman’s critique of QE is that it is not a “credible promise to be irresponsible”, since central banks will reverse the effects of QE as soon as inflation is back up. But the credibility of TPLT overcomes this problem. Further to its credit, a TPLT is consistent with the long-run mandate for price stability, and remains effective even if it is only credible with financial markets and not households or firms24. Indeed, the Federal Reserve’s recent change to flexible average inflation targeting means that in a twist of FAIT, this has basically become US policy. I think this is a reasonable sound foundation for monetary policy - but we can go further. One common market monetarist suggestion is the idea of targeting the nominal GDP level. That means targeting a growth path for the nominal GDP, much like price level targeting aims at the growth path for inflation. Nominal GDP is the price level multiplied by real GDP - that is, the amount of dollars spent in the economy in a certain period. That means it is a good proxy for the perfect target of stabilisation policy, which is some sort of weighted average of inflation and the output gap. But unlike this optimal target, it has a much lower epistemic requirement - we don’t need to know the output gap in real time or figure out which price index is the most representative and salient. And compared to a price level target, a nGDPL target is more effective against AD shocks. Consider what happens when there is a negative shock and real GDP falls. Firstly, a nGDPL target means that inflation has to be countercyclical and must rise - as we know, higher inflation lowers the real interest rate and stimulates the economy. Secondly, the fact that it is a level target means that people will expect the central bank to correct for any changes in the nGDP in the future, reducing the incentive to cut back on spending right now. Thirdly, a rise in inflation means that the real debt burden faced by a borrower will be reduced - that is, real income losses when rGDP falls will be shared between the debtor and creditor. A fall in inflation when rGDP rises will lead to higher real debt payments to the lender, avoiding the current problem of holding nominal loans. This countercyclical inflation means that the redistributive problems of inflation from earlier are less obvious, providing greater financial stability. Fourthly, a nGDPL target, by virtue of being a level target, provides the same benefits as a PLT/TPLT in terms of overcoming time inconsistency25. Consider what happens when there is a negative AS shock. Stabilising the price level forces all the burden of the shock onto a reduction in output. By contrast, stabilising nGDPL is more forgiving, allowing for some increase in inflation and some decrease in output. (My terrible MS Paint diagram is below.) In practice, flexible inflation targeting central banks are likely to allow the burden to be split - indeed, the flattening Phillips Curve would suggest that we should allow inflation to take on more of the shock than output, whereas nGDPLT forces it to be a one-for-one tradeoff. But while a perfect discretionary decision may yield better results than the rule-based option of the NGDPLT, the discretion of central bankers in practice has been worse than if we had just taken this rule as gospel. In part, the historical mandate of only looking at inflation has made central bankers less concerned with output than the general public. But a bigger problem is the aforementioned difficulty in differentiating between AD and AS shocks in real time - for example, the Federal Reserve kept interest rates low during the early 2000s, because a productivity boom caused a disinflationary AS shock. Milton Friedman and Anna Schwartz’s A Monetary History of the United States is a good reminder that as bad as crises might be, pro-cyclical actions by central banks can make them so much worse. A nGDPLT’s inherently countercyclical nature prevents this risk by not requiring the central bank to distinguish between AD and AS shocks. There are nonetheless good and meaningful critiques of nGDPL targets. Most of these revolve around the idea of inflation versus nGDP. For example, some may argue that we have spent a long time building up the public’s understanding of inflation targets and anchoring their inflation expectations, while nGDP is a much less clear concept. Others may point out that nGDP data is much more subject to revision, due to the worse quality of data available compared to on prices. Yet another critique is that since nGDP closely follows rGDP in the long run and since we are unable to predict rGDP growth due to our lack of knowledge over how productivity growth will change, this may lead to the target becoming out of date quickly. And finally some may note that focusing on nGDP means we allow costly variations in inflation. Hence my proposal - a temporary nGDPL target in the vein of Bernanke’s TPLT. That is to say, central banks would treat it as an intermediate target when we begin to face serious recessionary concerns, while keeping it as a useful metric on the backburner in other times. This deals with all of the issues above. Firstly, it is much easier to sell the idea of increasing nominal incomes rather than raising prices during an inflation, making communication an easier prospect. Secondly, it being temporary means that people’s long-run inflation expectations will not be de-anchored - the Bank of Israel has basically engaged in nGDPL targeting while keeping inflation anchored. But even if these expectations were to rise, we know how to deal with rampant and runaway inflation, thanks to Paul Volcker. The pressing concern in a recession is not some ghost of the 1970s, but the immediate collapse in people’s standards of living. The Bank of Israel has basically engaged in nGDPL targeting while keeping inflation anchored. Thirdly, the quality of data is an uncomparative concern - a flexible inflation targeting central bank needs to know about inflation, real potential output and real current output. Here, we just need to know about nominal output, which is approximated well by nominal Gross Domestic Income, a figure that we have pretty good data on. Fourthly, the worries about our inability to predict the long-run trend of nGDP is not a problem because we are using it in a transitory rather than permanent way. Finally, the costs of inflation are lower than we previously assumed27 - and in a recession, these are outweighed by the more salient issue of previous debt contracts, price negotiations and wage negotiations being conducted on the basis of a much higher nGDP trend. This is especially since it seems to me that one important “sticky price” in recessions isn’t what we traditionally consider, such as prices of goods and services - but rather of nominal wage and debt contracts, thus making stabiising nominal income important. But perhaps most importantly, a temporary deployment of a non-traditional targeting regime provides a pretty strong signal to the public that a regime shift has occurred. It is commonly known that the departure from the gold standard was one of the key reasons we managed to escape the Great Depression in the 1930s - but the specific transmission mechanism is less clear. In my Personal Investigation for the Pre-U History exam, I posited that this was such a large change in the monetary regime that it reset people’s expectations of inflation28, echoing recent research on the importance of credible policy regimes29. But don’t take my word for it - the former Chair of the Council of Economic Advisors Christina Romer argued that regime change was one of the most powerful tools in the playbook and was crucial towards the recovery from the Great Depression30. She has therefore pushed for “a very strong change in policy … a regime shift”, in order to get out of “a recession at the zero lower bound”31. Because of its inherently countercyclical guardrails, its in-built dampening of AS shocks and its capacity to provide a powerful boost at the effective lower bound, we should deploy a temporary nGDPL target when faced with a serious recessionary threat. Where does fiscal policy go from here? Let’s look at fiscal policy - this is much more straightforward. Given the constraints limiting fiscal stimulus as elucidated earlier, we need something that bypasses the inside lag (of the time between a shock and implementation) and the outside lag (of the time between implementation and an effect on the economy), as well as being targeted to those who are most vulnerable and likely to spend in a recession. That means implementing a trigger-dependent automatic stabiliser to provide direct cash transfers. Basically, this means that when a certain macroeconomic indicator meets some criterion, we will start sending money directly to everyone. To be more specific, a useful indicator is the Sahm Rule - developed by former Federal Reserve consumer section chief Claudia Sahm, it says that a reliable early warning sign of a recession is if the 3 month average unemployment rate rises 0.5 percentage points above its low point in the previous 12 months. This has been the case since the 1970s, and is a useful trigger for sending direct payments. It is pretty obvious to see that this is going to get money into the economy more effectively and faster than pretty much anything else32. This is undoubtedly imperfectly targeted and somewhat wasteful - but it has become clear that the general tendency in crises is for the perfect to become the enemy of the good, and optimal targeting is less important than a fast and feasible way of getting money to those who need it. As such, this represents a useful tool at the start of recessions when legislatures are still dithering over the details of their discretionary stimulus package. How can we improve our stabilisation infrastructure? For those who subscribe to the newsletter, you will have seen me mention the FPUC once every fortnight - that was the extra$600 a week in unemployment benefits the US government was providing up until the end of July. You may have even seen the campaign of #Savethe600. One of the most apparent difficulties was that even though it solely gave money to those who were getting unemployment insurance, for which there is already some sort of payments infrastructure, the money took forever to get to people.

To go on a tangent, think back to 2008 and the immediate aftermath of the financial crisis. Even as the Bank of England’s base rate fell, many banks did not lower the interest rates they charged consumers in a commensurate fashion. Why? Because they were in the process of balancing the book and deleveraging.

In both these cases, the transmission mechanism of either fiscal or monetary policy was impeded. We can fix this by investing in an improved payments infrastructure - specifically, a central bank savings account. The ability to open a fairly simple savings account with the central bank has several impacts. Firstly, it means that everyone has a bank account and no one is prevented from having banking services on the basis of high costs. This means that when it comes to a direct cash transfer, the government is able to do so on a universal basis, whereas government data on people’s bank accounts is actually surprisingly sparse right now. Secondly, most central banks have some sort of real-time payment systems - that makes the speed of transfer much easier. Thirdly, it means that the transmission mechanism for an interest rate change is much faster, since it does not rely on commercial banks to pass it over. By providing universal banking access, we can improve the efficacy of the two pillars of our stabilisation policy.

What is the state of stabilisation policy?

Let me end on a more positive note, because the scale of the reforms suggested make our economic outlook seem rather dreary. Bob Lucas famously stated in a 2003 speech that “the central problem of depression prevention has been solved”. That may seem absurd, given our current economic straits - but he wasn’t wrong. The 2008 financial crisis wiped out 5 times as much household wealth as the Great Depression, and yet the peak-to-trough change in GDP was one-fifth as large and job losses were less than half as large. We stopped a depression. And although it was imperfect, stabilisation policy is designed to stabilise - it isn’t there to replace the necessary process of deleveraging and rebalancing production within the economy.

In the 12 years since 2008, we’ve seen an incredible amount of innovation and creativity in stabilisation policy - the consequence of which is a bolder monetary and fiscal response in 2020. So we have improved and iterated in many meaningful ways. But the result of these two “once-in-a-lifetime” scale crises in such quick succession is that our arsenal is getting worn down. Even if we can prevent depressions in the future, the human and economic costs of prolonged recessions are significant - and that is precisely why we need to keep bolstering our recession-fighting capabilities with a temporary nGDPLT, a Sahm Rule automatic stabiliser and universal central bank savings accounts.