The second COVID relief package has finally been signed as of this writing, with questions about the size of the stimulus payments nearly derailing the relief effort— whether it should be 600$ or 2000$. The more important question to me is how effective stimulus payments are in actually creating stimulus (i.e. boosting consumption) and whether the economy as a whole would be better off if the stimulus is bigger or more targeted. Consumers can do one of three things with stimulus payments — spend it on consumption, save it for the future or pay down their existing debt, which is also equivalent to saving for the future (foregoing current consumption to be able to consume more tomorrow). There are three broad conceptual frameworks to think about consumer behaviour in response to stimulus payments.
Ricardian Equivalence
The economic theory of how individuals decide to consume is given by the Consumption Euler Equation. The idea behind the Euler Equation is that an individual weighs the costs and benefits of current consumption vs future consumption. For example, if an individual consumes more today, then he is setting aside less of today’s income towards savings. So he is forgoing the returns from those savings, which could have helped him consume more tomorrow. Based on this tradeoff individuals decide how much they want to consume today. In some cases, that means saving some money today and in other cases, it means borrowing money today to attain the desired consumption level.
One proposition stemming from this optimization behaviour of individuals (and this is not without controversy) is the concept of Ricardian Equivalence, i.e. if individuals know that a tax cut or stimulus payment they receive today is only temporary and the government is going to raise their taxes tomorrow to fill the budget hole, then they are more likely to save the money given by the government rather than spend it, i.e. it would have no economic impact. The assumption being made here is that consumers are “rational and forward-looking and want to smooth consumption over their lifetimes”. One key exception to this proposition is when individuals face “liquidity constraints”. If an individual’s desired consumption > actual consumption and he finds himself in a circumstance where he cannot borrow money to make up for the consumption shortfall due to a very high cost of borrowing, then he is more likely to spend any tax cut or stimulus payment from the government in a desire to achieve desired consumption. Economists also like to state this as: “liquidity constrained households have a higher marginal propensity to consume”. Poor people or people without jobs are more likely to be liquidity constrained and therefore, most fiscal stimulus is ideally targeted towards unemployed/low-income households as it is likely to have the biggest bang for the buck.
Savers-Spenders Theory or Rule-of-Thumb Consumption
Mankiw (2000) summarizes this behaviour among consumers. According to this view, not all consumers exhibit Ricardian equivalence and forward-looking behaviour. A sizeable chunk exhibit the rule-of-thumb behaviour (hand-to-mouth behaviour) of consuming current income. There can be several reasons:
- Consumers can face liquidity constraints of the kind explained in the prior section
- Consumers give too much weight to current income and sometimes naively extrapolate it to future income, i.e. they mistake a temporary boost for a permanent boost
- Individuals can have zero net worth, i.e.they do not have any savings to smooth their consumption over time and can even have negative net worth due to excessive credit card debt or other liabilities over financial assets
- Kaplan, Violante and Weidner (2014) add another type of consumer exhibiting hand-to-mouth behaviour. According to them, there are also “wealthy hand-to-mouth” consumers who have much of their wealth saved in illiquid assets (i.e. there is a significant transaction cost involved in adding or removing savings from such assets) and have very little liquid assets. So, it may be more optimal for them to spend transitory gains in income rather than save to smooth consumption
Mental Accounting Framework of Consumption
Thaler (1990) proposes that households think in terms of a system of mental accounts — a current income account, C (with a marginal propensity to consume close to 1), a future income account, F (with an MPC close 0) and an asset account, A (with an MPC between 0 and 1). The following exhibit from the paper beautifully summarizes his view on the consumption behaviour of households.
Several economists have studied direct stimulus payments closely over the last couple of decades and have some fascinating findings. Most of this research is based on survey data so the results should be taken with a pinch of salt. The 1200$ economic impact payment of The CARES Act was not the first such stimulus payment. There have been at least 4 other such payments in the past from 1992, 2001, 2008 and 2009.
1992 tax withholding changes:
In response to a recession since 1990, a tax policy change was initiated in 1992 so that a typical married worker gets a monthly take-home pay $29 higher, which seems low, but in aggregate amounted to 0.5% of GDP. However, it was explicitly laid out that the change was temporary and there would be lower tax refunds in 1993 as a result. This a classic example to test Ricardian Equivalence. With the full knowledge that the tax break is temporary, individuals should mostly save additional income rather than spend it. However, economists Shapiro and Slemrod (1993) conducted a survey after the change was implemented and found that nearly 43% of respondents indicating that they plan to consume the extra pay. Their analysis also found no strong relationship between liquidity constraints and the propensity to consume the extra income.
2001 tax rebate:
In 2001, a tax rebate of up to 300$ for individuals and 600$ for a couple was given in the form of a check dispatched between July and September 2001. The rebate was the result of a new income tax bracket of 10% created for 2001, which was earlier taxed at 15%. The 2001 tax rebate was considered less progressive because even higher-income households were eligible for it, for the part of their income that falls in the 10–15% bracket.
Economists Shapiro and Slemrod (2002) studied this stimulus as well through a survey. They found that the stimulus was substantial by several measures — around 92 million households got a check, with 72 million receiving the full amount, the rebates amounted to 0.4% of 2001 GDP and 1.5% of median income. They found that the spending rate was lower than expected with around 22% to 27% stating they were likely to spend it. However, they found that a significant % of people (46% to 48%) indicated they would use that money to pay down debt with the rest indicating they would mostly save it.
Johnson, Parker and Souleles (2005) conducted a separate study using survey data and found that in fact 20–40% of the rebate was spent on nondurable goods in the three month period when the rebate was received and about 2/3rds of the rebate was spent by the end of 3 more months. They also found evidence of higher spending among liquidity constrained households (low income or low liquid wealth).
2008 and 2009 stimulus payments:
The 2008 stimulus payments were the earliest to be distributed through electronic transfer in a big way (~40%), ensuring faster disbursement of benefits. These payments were tax rebates to the tune of 300$-600$ for individuals, 600$-1200$ for couples and an additional 300$ for those with dependent children. These benefits reached around 120 million households and were more progressive than the 2001 rebates because the benefits were phased out with higher income. It amounted to around 2.2% of 2008 GDP.
The 2009 stimulus payments were delivered in the form of a Making Work Pay (MWP) tax credit over two years. The credit amounted to 6.5% of median earnings, with rebates of up to 400$ for individuals and 800$ for couples and phased out for higher-income earners. This wasn’t a lump sum payment like the 2008 stimulus. For a weekly paycheck, individuals saw 10–15$ more. For those not working, such as social security recipients, a one-time transfer of around 250$ was made.
Parker, Souleles, Johnson and McClelland (2013) studied the 2008 stimulus and observed that survey respondents spent on average 50–90% of the payments in the quarter of receipt with 12 to 30% towards non-durables and services and the rest towards durable goods. They also find evidence for larger spending among low-income households and older households, and a stronger spending response of homeowners than renters. They find a similar response to stimulus payments whether by check or by electronic fund transfer (EFT).
Sahm, Shapiro and Slemrod (2012) evaluate 2008 and 2009 stimulus payments impact on consumption as well as the impact of means of delivery. The questions they address are:
- Does it matter if it’s a check or EFT — the hypothesis being the act of depositing a check is active while EFT is passive and it could affect consumption behaviour
- Does it matter if it’s a one-time stimulus payment or reduced withholding — the hypothesis being that small periodic gains should result in more consumption by the mental models framework as explained by the exhibit below from the paper:
Overall, their surveys found that over 75% of the payments are saved or used to pay down debt for both stimulus programs, with less consumption for the 2009 reduced withholding than the 2008 stimulus payment contradicting Schwarz’ hypothesis. However, they also admit that the 2009 withholding reduction benefit received very little publicity with many respondents being unaware that they received a benefit, while the 2008 stimulus payment was widely publicized by the media and the government. Sahm, Shapiro and Slemrod in an earlier paper, found no significant differences on consumption effects between EFT vs check.
2020 Cares Act Stimulus Payments:
As part of the Cares Act, economic impact payments to the tune of 1200$ for individuals and 500$ extra per child were given, with payments phasing out for higher-income households. The current recession is unique in the sense that consumption is constrained partly due to economic hardship and partly due to forced consumption suppression from lockdowns.
Economists at the New York Fed have examined how households used their stimulus payments using survey data. They find that by the end of June 2020, 29% was used for consumption, 36% saved and 35% used to pay down debt. They suspect that high uncertainty regarding the extent and duration of the downturn and social-distancing requirements could have suppressed consumption. They also find that payments spent on essential goods increase with age and decreases with income, while the reverse is true for non-essential goods.
Coibion, Gorodnichenko and Weber (2020) conducted another analysis based on a survey from July 2020. Their study reveals a similar skew towards savings with only 15% of respondents spending most of their stimulus checks with 52% mostly paying down debt and 33% mostly saving it. Per stimulus payment, the average household mix was 40% spending, 30% saving and 30% debt payment. However, they found significant heterogeneity with as many households spending all of their checks as there were households saving or paying down debt using all of their checks. As observed in earlier studies, they found more spending among liquidity constrained households.
Sahm, Shapiro and Slemrod (2020) also studied the CARES Act payments and found that only 20% mostly spend it with the rest either saving it or paying down debt. They estimate that at least some of the respondents who say they will mostly save or mostly pay down debt will increase their spending in subsequent quarters (consistent with earlier studies). Furthermore, they find significant differences in the ability to save/pay down debt by income, job status and by what race they belong to. A study looking at JP Morgan chase customers through May 2020 draws similar conclusions. They find that card spending, which collapsed 40% by the end of March, began to recover by mid-April across the income spectrum and with the largest improvements among the bottom 25%.
From all of these studies, we can arrive at the following conclusions:
- Most stimulus payments are either saved or used to pay down debt with around a quarter typically going towards consumption in the quarter of the stimulus
- However, there are significant differences in spending among households with plenty of evidence that liquidity constrained households overwhelmingly benefit from stimulus payments and mostly spend it
- Among the ones who save or pay down debt, there are lagged effects and it boosts their consumption in subsequent quarters
There is a case to be made therefore for larger stimulus payments, IF they can be more targeted and if the goal is to boost consumption and prevent a demand collapse. However, this time is clearly different and the pandemic induced lockdowns do suppress consumption. Saving is a good thing in this crisis because moratoriums on rent, loan repayments and credit card forbearance are eventually going to expire and in the absence of savings, it could lead to a consumption collapse coming out of the pandemic.
