Do people spend?

PART I: EXPERIMENT IN MACROECONOMICS

Let’s begin with a digression….

Randomized Control Trial (RCT) is now a household phenomenon. Everyone knows something about it. The 2019 Nobel Prize in Economics to Kremer, Banerjee and Duflo is the inflexion point of this movement. The Trio contributed immensely to the public discourse and policy by applying RCT in development economics. The Nobel committee recognized their contribution on ‘experimental approach to alleviating global poverty’.

Seldom in billboard you see academics and especially social scientists. In my trip to Kolkata, I saw how the city had put billboards welcoming Prof. Abhijit Vinayak Banerjee. Well, it was a proud moment for everyone around the world with some involvement in development aspect. It was a proud moment for the subcontinent and of course for Bengalis. After all Abhijit is the fourth Bengali to receive this prestigious award from this subcontinent (the other three are – Tagore in 1913, Amartya Sen in 1998 and Muhammad Yunus in 2006).

There are challenges in RCTs and that is for another day another topic.

While microeconomics offers the opportunity to conduct controlled experiment (not so controlled like your science lab), it is not so in case of Macroeconomics. In macroeconomics, the researchers have very little control over things the way they unfold. Also, it is difficult for government even to adopt this experimental approach. Can the central bank set two different rates for two different states to see what is working well? Similarly, can govt. have two different tax rates for people fulfilling same eligibility conditions to see if there are spatial differential impact? Seemingly impossible.

Does that mean macroeconomics is not subject to any experiment? It is and if you want to oversimplify it, you can say it is by the act of GOD. We call it technically as Natural Experiment in which treatment and control are determined by nature or factors beyond researchers control.

Nicola Fuchs-Schuendeln & Tarek Alexander Hassan (NBER, 2016) have written a 100 page review on Natural experiments in Macroeconomics. They have covered topics such as permanent income hypothesis, fiscal multiplier, causal factors in economic growth etc. Highly recommended for students who want to understand how various events have intended and unintended consequences and how they unconsciously help us to understand many unexplained phenomena.

One such study which probed an interesting question was published in 2009 in American Economic Review – an authoritative journal in economics. The study is titled as ‘Did the 2008 Tax Rebates Stimulate Spending?’ authored by Matthew D. Shapiro and Joel Slemrod – both from University of Michigan. They have probed the question whether people have responded by spending the additional money they obtained through the US tax rebate of 2008? This will help us to understand how the fiscal policy may work – producing at times completely unintended consequences. 

PART II: Did they spend?

Remember a couple of things before reading rest of the post:

• Permanent Income Hypothesis – Your consumption today is not only the function of today’s income but also by your future expected income. If you have an odd-even formula in your company that in the even month you get Rs. 50,000 and in odd month you get Rs. 100,000, isn’t it logical that you smoothen your consumption over the entire 12 months?
• Marginal Propensity to consume (MPC) by the poor is high – Have you read something like that? Poor people do spend more of every additional rupee they receive vis-à-vis rich people. When govt. plans for more public spending, economists give them a back of the envelope calculation saying – well, given that the average MPC is this, there would be a delta rise in aggregate income in the country applying the usual Keynesian multiplier. 

Here goes the story from the paper of Shapiro and Slemrod -

On February 13, 2008, US government had enacted Economic Stimulus Act of 2008 to stop the economy from sliding further. The total fiscal cost to the exchequer was to the extent of USD 152 billion of which two-third was meant for tax rebate to be provided to 130 million households. The tax rebate was something of this nature –

“Most taxpayers below the income limit received a rebate of at least $300 per person ($600 for married couples filing jointly). Eligible taxpayers received, along with their individual payment, $300 per dependent child under the age of 17. …..”.

[Source: Look at Wikipedia entry for details– Economic Stimulus Act of 2008 as well as cross references]

What was the rationale for this stimulus package? It was pushed as a recovery rebate and an incentive for business investment. It was expected that individual tax rebates would provide incentives to consumer to spend, and such spending would result in demand for goods and services and give boost to business.

What was the net result? Most of the money was either saved or used to pay off the old debt. Very little was spent. This is called unintended consequences. This is where Economists add value. While Rawls and Nozick do matter, but as Macroeconomists, they can produce the evidence of action and give a logic of why and what had happened. It is for the elected representatives to take the call on the margin on the basis of these learnings.

Shapiro and Slemrod used the survey data of University of Michigan. University of Michigan conducts a monthly survey known as Survey of Consumers. The authors used this vehicle to add a few additional questions to understand the dynamics of this tax rebate. This is what the result looks like:

This is what you would not expect. Right? Look at Figure-1. Only one-fifth of the respondents said that they will ‘mostly spend’. Else a whopping 80% have said either they would use the money to pay-off the debt or save. Nearly 50% said they would use the money to settle the past debt. This is also quite revealing.

Households do really care about their balance sheet and during a crisis phase when their balance sheet get impaired, they would rather repair that instead of getting into new spending. Poor people know that they are liquidity constrained and they know that they have a bad ‘today’. Also, they know that if they have to keep the social collateral intact, to fall back upon during a crisis, they have to preserve that trust. They would rather return the money they borrowed in the initial few months than spend any additional money they would have received.

This is where the commonsense fails and you may better rely on past evidences. This is also a situation which produces ethical dilemma – do I really care about the poor and better give cash in their hand or should I do something which would revive the economy and ultimately help the poor? If you give cash in hand, given the balance sheet situation of households, they might use it either to pay off the debt or save it for tomorrow. If they do so, your dream of reviving the economy through Keynesian channel would be far longer, if not absolutely a failure. This is a tough call.

What is the pattern of saving by age? Look at Table 2. No, it is not what we might have thought about – young people spend more than the old people. If you compare people with the age of 29 or less vis-à-vis people of age 65 or more, you could see a larger fraction of people in the older age bracket are ‘mostly spending’. Why? People do plan for their consumption and spending depending upon their age or we can say it as per their life cycle. If they have to have a stable lifestyle, young people have to save more as there is a prolonged tomorrow for them than the old people.

Is it universally true?

There is nothing universally true in economics. Is it largely true? It depends upon the context – time, society, sub-population category etc. Then what’s the fun of having these grand statements? Well, it can only help you to set a hypothesis. When you do not have the data and have to do mental arithmetic (invoking your system I), these can be helpful.  

Do ‘not so rich people’ spend more than the relatively rich people? Remember the MPC point we had given in the beginning of this section.

The paper says:

Congressional Budget Office (2008, 7), in its analysis of the options for stimulus in early 2008, states, “Lower-income households are more likely to be credit constrained and more likely to be among those with the highest propensity to spend. Therefore, policies aimed at lower-income households tend to have greater “Lower-income households are more likely to be credit constrained and more likely to be among those with the highest propensity to spend. Therefore, policies aimed at lower-income households tend to have greater stimulative effects.

Look at Table-3 above. There is nothing you can conclude out of it. If you just take the low-income household (Income <USD 20,000) and high-income household (income>USD 75,000), more people in the higher income bracket are willing to “mostly spend”. The paper concludes – there is no discernible difference in spending propensity by income. However, more people in the income category of less than USD 20,000 are inclined to use the money to set off their debt.

In conclusion:

This is a natural experiment. The researchers have just piggybacked on the University survey to uncover various interesting findings. Is it credible? Well, this study got published in American Economic Review and NBER. However, unless there is a systematic review or meta-analysis, its very difficult to say that if it holds true more often than not.

The paper however gives some conjectural evidence or circumstantial evidence. Let’s look at this figure.

As you could see, personal saving (expressed as % of disposable income) has increased by the same amount as the rebate. The paper says – “personal saving jumped by nearly as much as the increase in the stimulus payments provides striking circumstantial evidence for little contemporaneous spending from the rebates”.

 It also says – “Because we do not know the counterfactual—what saving would have been absent the rebate—the figure does not establish what fraction of the rebates was spent”.

What a beautiful way to portray the ‘as-is’. There could not have been any better framing than this.

It is important to be honest and candid in claims in Macroeconomics. Else, a lot of circumstantial evidence would be pushed as causal evidence. While Macroeconomics does not provide a lot of opportunity to do controlled experiments, such natural experiments are quite helpful to draw lessons for the future.