The American Rescue Plan: A lesson on evidence-based policy debate
Leading macroeconomists in the US recently debated the likely impact of President Joe Biden’s American Rescue Plan (ARP). If you follow this debate you will learn a lot about the ARP. But that is just an “entree” specific to the US. The more generalisable main course is how macroeconomic theory and evidence is used by economists who, like most human beings, have two hands. Consider the discussion below as a case study on the nuances of macroeconomic policy discourse counting dollars and cents to lift an advanced economy out of recession faster than it otherwise would
The ARP, passed on March 10 by the US Congress, is a $1.9 trillion (nearly six times Bangladesh's annual GDP) emergency fiscal package with health, social protection, and business support measures to fight the Covid-19 pandemic and its impact on families and communities in the US. It intends to hasten the end of the plague; cushion people hurt financially by the pandemic; and provide aid to those who need help but do not qualify under tightly targeted programs (described as belt-and-suspenders by Krugman).
The most vocal support for the package has come from Paul Krugman, Treasury Secretary Janet Yellen, Fed (US central bank) Chairman Jerome Powell, and the IMF's Chief Economist Gita Gopinath. They believe the package is just about right, given the uncertainties, with manageable downside risks.
Larry Summers and Olivier Blanchard are vastly more circumspect. They think "stimulus measures of the magnitude contemplated are steps into the unknown". A macroeconomic stimulus on such a scale will set off inflationary pressures with several (unintended) consequences for the value of the dollar and financial stability.
There is general agreement among both camps that such a package is needed. They disagree on the appropriate size and the severity of downside risks.
Conceptual underpinnings
A short pedantic detour into a few basic concepts may be useful for some readers.
The fiscal package, technically, is responding to a problem measured by the Output Gap – the difference between the economy's current level of output, empirically the Gross Domestic Product (GDP), and the level the economy normally produces (potential output) when non-labor resources are fully employed, and unemployment is around the lowest level (the natural rate) that the economy can sustain without creating inflation.
The output gap is negative in a recession, as is the case now, and positive when the economy is expanding too much. The deviations of the actual from the potential are never static. Dynamic theoretical models explore the conditions that can make the deviations implode or explode over time. Anything (multiple equilibria) is possible in a dynamic system of interdependent non-linear relations. Economies can take too long to crawl out of recession, grow destabilising bubbles, and crash land from inflation. Moderation of the output gap provides the rationale for stabilisation policies.
Gauging the size of the intervention needed to eliminate the output gap requires knowledge of the Multiplier. This is the change in output due to an exogenous change in expenditure such as a fiscal package. The spending by the direct beneficiaries boosts the incomes of others which they in turn spend and so on. The resulting increase in real (inflation-adjusted) GDP for each initial dollar of public spending is called the fiscal multiplier. The size and time path of the output response to any policy impulse depend on the multipliers. They are variable across types of expenditures, access to liquidity, and the state of the economy.
How the impact on output translates into employment is ascertained through the Okun's Law. This is a statistical correlation describing the percentage by which unemployment falls when GDP increases or vice-versa. This percentage is the Okun coefficient. Output depends on the amount of labor used in the production process, so there is a positive relationship between output and employment. Total employment equals the labor force minus the unemployed, so there is a negative relationship between output and unemployment.
The value of the Okun coefficient widely used in the US is 2. A 1% rise in unemployment is associated with 2% fall in GDP. Conversely, when GDP rises by 1%, unemployment falls by 0.5%. Okun coefficient is generally lower during expansions and higher during recessions.
How unemployment relates to inflation is described by another statistical correlation called the Phillips Curve. A higher inflation rate is associated with a lower unemployment rate and vice-versa. High demand in a booming economy provokes workers to seek higher wages and firms to raise prices. Inflation is generally sensitive to the unemployment gap – the deviation of actual from the natural unemployment rate. The widely used Phillips Coefficient currently is between 0.2 to 0.5 – one percentage point decrease in the unemployment gap goes with roughly 0.2 to 0.5 percentage point increase in inflation over time.
The stability of the Phillips Curve depends on long-term inflation expectations, i.e. the rate at which consumers, businesses, investors expect prices to rise in the future. Actual inflation generally depends on what we expect it to be in addition to excess demand and cost shocks. How strongly inflation expectation is anchored, meaning relatively insensitive to incoming data, is critical.
Inflation expectations are well anchored when the public experiences a spell of inflation higher than their long-run expectation, but their long-run expectation of inflation does not change as a result (a la Ben Bernanke). They are weakly anchored when long-run expectations react to a short period of higher-than-expected inflation. Beliefs about the long-run behavior of the monetary authority and ultimately the political process around it shape the long-run inflationary expectations.
Return now to the crux of the debate: Is the size appropriate? How much should one worry about the downside risks?
The stimulus is oversized
Best to begin from a consensus point. The ARP is bigger than the output gap, no matter how you cut it.
Krugman takes the annualised fourth quarter output in 2020 as potential to conclude the US economy is 4.3% below where it could be, implying a $950 billion shortfall. This is roughly equal to income support under ARP. Adding pandemic control and belt-and-suspenders makes the package bigger than the decline in GDP.
Larry Summers argues wage and salary incomes are about $30 billion a month below pre-covid-19 forecast. This gap will likely decline during 2021. At about $150 billion a month, income support is at least three times the size of the initial output shortfall. Relative to either the output gap or declines in family incomes, the package is exceptionally large.
Olivier Blanchard stepped in with arithmetic simple enough to tweet. He uses the 3.5% unemployment rate before Covid-19 as the natural rate. The Congressional Budget Office estimated the potential real growth for the past few years at around 1.7%. Actual real GDP in the last quarter of 2020 was 2.5% below its level a year earlier. This implies an output gap in 2020 4th quarter of 1.7 + 2.5 = 4.2%, equivalent to about $900 billion, not vastly different from Krugman's $950 billion estimate.
There is therefore no question ARP is oversized. The proponents say the risk of doing too much is less than the risk of doing too little while the dissenters find the size "much too much".
Do not worry, sleep easy
Not all components are expansionary, argues Krugman. Compensating income losses does not add fresh demand. For many of the belt-and-suspenders recipients the checks will add to demand, but the multiplier is likely to be low, "because people who didn't need the checks will largely save them." Pandemic control expenditures may have a reasonably high multiplier.
A 5% excess of actual over potential output implies an unemployment rate about 2.5 percentage points below the natural rate. Assuming an effect on inflation of about 0.2% for every 1 percentage point decrease in the unemployment rate, inflation would increase by 0.5%. Even taking larger estimates of the slope coefficients, such as those around 0.5, the increase in inflation would be around 1.25%, hardly anything to lose sleep over when expectations are expected to not react to movements in actual inflation.
Gita Gopinath of IMF explains why expectations may not react. Despite large swings in the US unemployment rate from 10% in 2009 to 3.5% in 2019, inflation remained remarkably stable around the targets set by central banks, even as wages rose. Globalisation limited inflation in traded goods and some services; automation, along with relative declines in the price of capital goods, largely kept higher wages from being passed through to prices; and the dominance of market share by firms with high profit margins allowed these firms to absorb higher costs without raising prices.
Global supply chains have shown resilience and agility during the pandemic. Considerable slack remains in the global economy. Automation trend is likely to accelerate. Market share of firms with high profit margins has increased as the pandemic-related downturn has affected smaller firms harder than large businesses. Monetary policy is unlikely to target higher inflation because the political tolerance is low.
The risks are asymmetric. If the economy is weaker than expected, the Fed may not be able to offset it because monetary policy has little room left. If it is stronger, tighter monetary policy is feasible. So, it makes perfect sense for the package to be substantially bigger than the "output gap". The loss from spending too little would most likely be concentrated among people already hit the hardest.
The Fed is not expected to react to the higher demand or increased inflationary pressures by raising interest rates if inflation rises by a couple of percentage points from the 1.7% in February 2021. It is unlikely to tighten monetary policy given their new flexible average inflation targeting framework. Indeed, the Fed has signaled that faster inflation and inflation temporarily above its 2% target would be welcome. Markets have faith in central banks' independence and their policies are credible.
Why go there?
Summers and Blanchard are not so sanguine. An overheated economy could set off inflationary pressures "not seen in a generation". How so?
Unemployment is falling, monetary conditions are loose, and there is likely to be further strengthening of demand. With progress towards herd immunity, consumers are poised to spend down the approximately $1.5 trillion which they accumulated from pandemic curtailed ability to spend in 2020. They will spend some fraction, say half. Coronavirus Aid, Relief and Economic Security added $900 billion of stimulus in December. ARP is $1.9 trillion. This adds to a $3.6 trillion impulse to demand, four times the $900 billion output gap.
Blanchard figures if the average multiplier is 1, demand would increase by $3.6 trillion, leading to a level of output 14% above potential, thus taking the unemployment rate close to zero. Inflation expectations may rise sharply from the current non threatening levels. "Overheating may turn into a fire!"
If the Fed lets inflation increase substantially, inflation expectations would likely become de-anchored, warns Blanchard. This will make monetary policy more difficult to use in the future as well. If the Fed tightens monetary policy substantially, the increase in interest rates might have to be large, leading to problems in financial markets and a crash landing.
Summers contends containing an inflationary outbreak without triggering a recession is more difficult now than in the past. Every past significant inflation acceleration has been followed quickly by a recession. Taming inflation will require allowing unemployment to rise. Engineering a soft landing is difficult. The flatness of the Phillips curve implies that if inflation does rise, policymakers will have to accept substantial increases in unemployment to make it fall.
No smoking gun on either side
There is no certainty about what is going to happen. The unique features of the pandemic induced recession make evidence from past episodes on either side not entirely pertinent. A negative supply shock hit some parts of the economy but not others. The former created a negative demand shock for the latter. The economy has catapulted into recession because of the fear of a deadly disease. The impact of an oversized ARP on inflation in such a state of the economy is ambiguous in theory.
The pandemic decreases the size of the fiscal multiplier. Receiving a check during a pandemic will not necessarily make people spend it at restaurants and bars. Many may instead keep it in the bank, pay down debt, or settle rent arrears. Stimulus used for these purposes works like a retroactive social insurance, as pointed out by Noah Smith, that reshuffles financial claims and alters the distribution of wealth without creating additional demand.
Of course, many people will buy more with their government checks. Krugman admits that emergency spending not intended as stimulus may nonetheless have a stimulative effect. They may bid up the prices of consumer goods, hurting the people who did not get the checks. It could also be that a few idle resources are put to work, in which case the relief checks may stimulate output by preventing deflation.
Krugman's war parable makes the whole debate moot. "When you're fighting a war…you spend what you need to spend to win the war." The inflation risk is "an argument for seeking ways to limit that risk, not for skimping on Covid relief."