“Equitable Growth in Conversation” is a recurring series where we talk with economists and other social scientists to help us better understand whether and how economic inequality affects economic growth and stability.

[Editor’s note: This conversation took place on August 2, 2018.]

Heather Boushey: It’s so great to talk to you today. Thank you for making time. It’s—I’ve really been looking forward to this conversation.

Karen Dynan: Thanks. I have really been looking forward to this conversation as well.

Boushey: I want to ask you a lot of questions today about your work on consumption. You’ve done research talking about who buys what and how much, and how that affects our macroeconomy. And so, I want to sort of get right into this big question. Milton Friedman, a Chicago School economist, famously laid out the idea of what he called a Permanent Income Hypothesis, which is the idea that a person’s consumption is really only affected by their lifetime income. His hypothesis was that we all spend whatever we’re going to spend this month or this year based on our perceptions of how much we are going to earn over a lifetime.

When he came up with this theory in the 1960s, large datasets on consumption and income weren’t readily available to economists. Now, of course, we’ve had this huge increase in the amount of data available, a huge increase in computing power. Can you tell me a little bit about your research that gets at this question—about whether this “permanent income” idea is the way people spend—based on your use of new data, and empirical techniques and computing power that certainly weren’t available in his time?

Dynan: Well, let me start by saying I think the Permanent Income Hypothesis is a good benchmark for thinking about consumption. In fact, I taught consumer behavior to my students, and when I did that, it was the first model that I showed them.

The Permanent Income Hypothesis basically yields sensible predictions for a world in which people are patient, forward-looking, able to borrow freely, and good at doing complicated calculations about optimal consumption—and then also good at sticking with the plan. So, for all those reasons, I think it’s a good starting place in some senses, but, of course, the real world is much messier. That’s why we need to go to the data and also why these datasets that have information about individual households are so important to understanding consumption.

Anyone who has looked at the microdata knows that the Permanent Income Hypothesis doesn’t actually characterize the behavior of that many households. And this is something that comes out in my research, and that also has come out in an enormous body of research that is looking at this question. So, we see households consuming out of predictable changes in income. We see them consuming much more out of explicitly temporary changes in their income than the Permanent Income Hypothesis would predict. We see households nearing retirement that have accumulated almost no nest eggs to supplement their pension incomes. And we see a lot of variation across households in their behaviors, with some households having much higher propensities to consume out of income than others. All of these findings suggest that we need more complicated models, and maybe different models, to explain different people.

Boushey: So, tell me a little bit about what you’ve learned about the effect of differences in income and how that affects people’s consumption patterns. How does that fit in with the Permanent Income Hypothesis? And what has your research shown you about that?

Dynan: I’m going to speak kind of generally. I’m going to make statements about how people lower in the income distribution tend to behave, though it’s not true of everyone in the lower income distribution. Generally speaking, however, we find a higher propensity to consume in the lower part of the income distribution and also in the lower part of the wealth distribution.

This makes sense because we know this part of the population is most likely to be constrained, especially if they are consuming less than they would like to be consuming. So, when they get additional income, they consume a lot of it. This part of the population probably also struggles more when it comes to financial planning and is likely to lack access to good financial guidance, which means they have more trouble optimizing and so they tend to use simple rules of thumb.

That’s one general finding—that you find people lower in the income distribution having a higher propensity to consume than those who are higher in income destination. But I should add that when it comes to people who have high propensities to consume, it’s not that small a group, and it’s not just the people at the very bottom. You can go pretty high up the income distribution and still find people who are doing very little saving and whose consumption is very responsive to their income. Depending on the data and the time frame you are considering and what sort of consumption you are looking at, you can get numbers that are as high as half, or maybe even a greater fraction, of consumers who have a very high propensity to consume.

One very interesting finding in the data is that there is some evidence that there are actually wealthy hand-to-mouth consumers, people who are at the top of the income distribution who have high propensity to consume. I think this group is probably just too busy to be constantly reoptimizing, so they’re using these simple rules of thumb over the longer run. I think they also are probably high savers when it comes to the longer run. But certainly down at lower part of the income distribution, you’re finding many people with high propensities to consume.

Boushey: You’ve spent a lot of time both being a policymaker and advising policymakers here in Washington. Tell me a little bit about how you would explain to a policymaker why this research matters. How does this affect what somebody at the U.S. Treasury Department or the White House should think about economic policymaking?

Dynan: There are important implications for policy along several dimensions. One issue that we thought about a bunch when I was at the Treasury and when the U.S. economy was still weak in the wake of the Great Recession is that the propensities to consume matter when you are trying to design countercyclical economic policies. When the economy is in a slump, you want to target fiscal support to households with higher propensities to consume in order to provide the biggest boost to demand so as to get the economy back to full employment sooner. Likewise, you want to make sure that countercyclical monetary policy can reach people with high propensities to consume. This was an issue during the past recession because many people that probably had high propensities to consume were underwater with their mortgages and had difficulty refinancing their mortgage loans into lower-rate loans that would increase their discretionary monthly cash flow. So, monetary policy had more trouble reaching these high-propensity-to-consume consumers.

But there also are important longer-run policy issues. I think a key insight is that people with higher propensities to consume tend to struggle to save, which goes hand-in-hand with the idea that if you have a high propensity to consume, you can have a low propensity to save. These households end up with low levels of accumulated financial assets, and that can impede their mobility if they lack funds to, say, finance an education or start a business. So, policymakers need to keep this in mind when designing policy to increase opportunity and mobility.

It’s also important when it comes to retirement policy. Folks who end up in retirement with no accumulated assets may well find themselves facing hardship because their pension income is insufficient. That needs to be a consideration when it comes to evaluating potential changes in the Social Security system.

And when you look at the data, the numbers are really startling. I was just looking at the most recently released data from the Survey of Consumer Finances, and the levels of accumulated financial assets are so low, especially when looking in the bottom 40 percent of the income distribution. That’s concerning enough, but what’s even more concerning is that the amounts have trended down over time. I once might have said, “Well, we had the bubble, and, of course, they had temporarily high financial assets right before the crisis struck.” But if you go back further, if you compare today’s level of accumulated financial assets to the accumulated financial assets as of the year 2000, you can see that households have less than they did then. That finding is quite startling.

Boushey: I want to come back to that point in just a second, but I want to get in one more question about the amount of propensity to consume. Is there a good rule of thumb that you have about this concept? Is there a typical marginal propensity to consume, or is there a different one across the income distribution? Do you have a good rule of thumb there?

Dynan: I think it’s hard to have a single rule of thumb because the nature of the shock to income matters. Is it a temporary shock to income or a permanent shock to income? If it’s a permanent shock, then you would expect a high propensity to consume even for people much higher up the income distribution. But I think the propensity to consume is always going to be close to one for many of the households who are in the bottom half of the income distribution.

Boushey: Wow.

Dynan: Yeah.

Boushey: So, to connect the dots between what you’re saying about families not having a lot of assets and people spending most of what they earn, one thing that’s been talked about a lot over the past couple of years in macroeconomics and in the popular media is that the rich have a higher propensity to save. This might be contributing to the idea that income and wealth is concentrating at the very top of the distribution, which, in turn, is pushing down overall consumption, and which, in turn, is slowing down economic growth and acting as a source of financial instability. What’s your reaction to this idea?

Dynan: I find it quite plausible that saving would be lower and demand would be higher today in the aggregate if not for the increase in the income inequality that we’ve seen. So, the basic logic here is that growing income inequality has channeled more income to households that have low propensities to consume over time—those at the top of the income distribution. And we know that the growth of income inequality has been most pronounced at the very top of the income distribution, where propensities to consume are presumably quite low.

So, I find it very plausible that this is holding back aggregate demand, and, in turn, I find it quite plausible that it is one of the factors that has helped to drive down the equilibrium real interest rate over time. And that, in turn, has contributed to financial instability and created some very large problems for central banks in terms of the scope for using conventional monetary policy. I think if you just connect the dots between the findings in various literatures, it seems like that effect must be at work. But I think it’s an open question still as to just how important that channel is quantitatively. If I were to guess, it is one of several factors that collectively are making a noticeable difference.

Boushey: This is connecting the dots between a lot of different findings. Is there a way that somebody could be looking at that question directly? Have you seen research on that? Are you doing work on that? Or is that too difficult?

Dynan: We’re talking about the equilibrium global real interest rate, so it’s unlikely that precise findings will pop out of the microdata for you. But there are fairly straightforward calculations we can do because we know how much more income is going to people in different parts of the income distribution. You can then look at some of the evidence that we have on how the propensities to consume and to save vary across the income distribution. I think you can do these calculations.

I think the place where it would be great to get more evidence is on the variance in marginal propensities to consume. There’s been some interesting literature on this topic, including my own, but I think there’s more that we have to learn about these differences, particularly since so much of the action is going to be at the very top of the income distribution. That’s where the big amounts of money are, and we just have very limited microdata to look at those people.

Boushey: To recap what you’ve just said, we need more evidence on how consumption differs across the income distribution or across the wealth distribution. And you said variance, so it could be differences across people. But it’s hard to access the data because so much of the rise in inequality is at the top. Is that a fair summary?

Dynan: Yes. The point here is that only some of our datasets are useful for looking at what’s going on in terms of income at the very top. Tax records are in that regard, and we have the Survey of Consumer Finances, which oversamples wealthy households in order to kind of understand what’s going on with the wealth they’ve accumulated. But those datasets just don’t have the data on changes in consumption and changes in income that you would need to calculate a marginal propensity to consume.

Boushey: I don’t even know how you would get that kind of data unless you had access to income or credit card data. That’s the big data question.

I have one more question for you. As you know, Equitable Growth has a competitive grants program that funds academic research. Given the body of work that we’ve talked about today, where would you advise us to be looking for big questions that you think need to be answered? You’ve said one about more evidence on the variance of the marginal propensity to consume, but what else jumps out to you?

Dynan: I’m also interested more broadly in household economic security. I view all these things as linked very closely. My research is on how the new economy is affecting households. We all know the nature of work is changing rapidly, and there’s been some excellent research documenting just how many workers have shifted into alternative work arrangements such as contract work or gig economy jobs. And there’s been a lot of speculation about the downsides of these arrangements such as the loss of traditional worker benefits and traditional protections. But we don’t have a lot of hard evidence on how the economic security of these workers has actually been affected.

So, for example, one could make an argument that, on the one hand, this trend means income volatility has risen because more work is on-call, such that workers have much less control over their hours. On the other hand, many gig economy jobs allow workers lots of control over their work hours, and therefore they should have a greater ability to buffer unexpected spending needs. I think, broadly speaking, it would be great to get more evidence on this question and then link it back to what we’re seeing in terms of households’ ability to save. It looks like households are struggling more to save, but is this something that’s a mitigating factor?

Boushey: Those are big and important questions, and I love the way you connected the dots between the changes in the structure and nature of work and the microeconomic questions about family economic security. That’s a nice place to end. Thank you very much, Karen. I really appreciate it.

Dynan: It was wonderful to talk to you, and I just can’t emphasize enough how glad I am that Equitable Growth is doing the work you are because you all are funding research that answers questions that are so important to the policymaking community.