One Election Takeaway: Voters Hate Temporary Safety Nets

Announcement: I'm putting on an end of year sale. For just 55 dollars a year you can subscribe and receive premium pieces of Notes on the Crises. This also helps support the FOIA efforts I've been engaging- like the 30,000 page database of Federal Reserve Board minutes I got through FOIA which I launched this week. If you like my work but my typical price is too expensive, now's the time to subscribe.

As everyone knows by now, Donald Trump is president. Again. I do not think it is any secret that I’m not a fan of Donald Trump. At the same time, I’ve generally tried to keep partisan political opinions in this newsletter to a minimum. I try to keep the “politics” of the newsletter squarely focused on policy, while providing broader economic analysis. The discussion of how the economy interacted with the election is a perfect opportunity for me to return to themes I covered in the first year of the newsletter.

One theme I regularly took up was the fact that congress responded to the Coronavirus pandemic with a series of economic measures which were powerful, but time limited. These covered the gamut, but the most important were the expansions and extensions to unemployment insurance —the direct economic payments (“checks”) and the extensions to medicaid. These programs were important not because they provided “fiscal stimulus”, but because they underpinned households' livelihoods at a difficult time and facilitated “social distancing”. My worry then was that having such large programs on a time limited basis created what I called “fiscal cliffication”. That means politics would increasingly revolve around large fights over what to do as big programs approached their expiration dates — or passed it. As I said in August 2020

While this is disastrous for the country at large, the political incentives each party faces are going to lead to intensifying fiscal cliffs for the foreseeable future. Congressional Republican incumbents benefit from instability among lower and middle income households. Voting is deeply tied to residency and housing, which is obviously disrupted by mass evictions. Meanwhile, Democrats can correctly point to Republican obstructionism to drive turnout from their preferred voters — affluent suburbanites.

I normally don’t wade into partisan politics like this, but it is important to get a handle on these dynamics to understand the future of economic policy for the next few years. It was easy to get lulled into a false sense of security by the bipartisanship in the CARES Act — but that was a rare exception. The stars aligned to give both parties an unusual incentive to accomplish an overarching deal quickly. That is now over.

The American Rescue Plan passed in March 2021, with all republicans voting against it. Of course, by that point discussions of inflation had begun. Inflation then became ever present as a topic, over the course of 2021. I had already begun to focus on this by March 2021, though I wasn’t able to write about it anywhere as much as I wanted to over the following year. 

On March 9th 2021 I sent out an interview I did with Joe Weisenthal for his newsletter with this introduction:

For a long time now I’ve been thinking about, and foreseeing, a widespread turn to concerns about inflation. Early on in the pandemic I was concerned that supply chain disruptions coming from the abrupt shift to producing in pandemic conditions would lead to an inadequate fiscal response to the pandemic out of inflation concerns. It turned out that those supply chain disruptions were less dire than many feared- but disconcertingly because workplaces tended to not close and let employees bear the burden in the form of illness and death. Meanwhile, debates over the latest Covid relief packages in congress have not really focused on inflation or the availability of physical resources (I’ll have more to say about the latest relief package in the future). Yet we have seen price increases in commodities whose prices are determined on international chartered exchanges and supply chain disruptions have still led to unprecedented delays in deliveries and accumulating unfilled orders.

It turns out that those commodity prices, along with rent increases, were the center of the growth in measured inflation.

As I had started to worry about in 2020, the supply chain disruptions and other economic dislocations caused by the rapid changes to production as a result of Covid had led to an abandonment of fiscal policy. Congress drove over those various fiscal cliffs under the assumption that low headline unemployment made these programs unnecessary, while “inflation” made them undesirable. There were also economists, employers and politicians — which included most republicans and many democrats — who openly advocated unwinding unemployment insurance support to compel lower wage workers “back to work”. As the conservative Democratic senator from West Virginia Joe Manchin summarized this attitude, he did not want the U.S. becoming an “entitlement-based society”.

Which brings us to this election. Joe Biden spent the year he was running for president running on his foreign policy record, as well as his economic record. He emphasized both the superior performance of the U.S. economy relative to other “G7” economies in terms of “real GDP growth”, and low headline unemployment. He also emphasized his record of deficit reduction- a point we will return to. What he did not do was emphasize the economic support he brought to households in 2021 — because nearly all of them would expire by the time he was running for president. The rest would expire over the coming year — student debt payments were even restarted this very September.

The introduction and then subsequent final expiration of these programs are an extremely underdiscussed element of the election results (though since I started working on this piece, some coverage has emerged). Yes, price increases, particularly those for food, energy and rental housing had a crucial- and continuing — impact on many households, particularly those at the bottom half of the distribution of income. But the income drop off from the expiration of these programs made it a brutal double whammy. It’s important to understand how much these programs reduced both income inequality in the United States, as well as the probability of experiencing income declines.

To understand this point, we must also understand that statistics which either track an individual person — or individual household’s — economic experiences through time directly are far superior to statistics which take multiple snapshots of a group’s experiences at a point in time. Think of it like musical chairs. If the labor market experience is like endlessly cycling through the first three rounds of musical chairs in an unpredictable sequence — where some chairs are far more uncomfortable than others — then only a minority will be without a chair, or in an uncomfortable chair, at the end of a round. But we can’t conclude from those snapshots that the vast majority of people are having a successful time playing musical chairs. This is even true if the chairs are overall improving in quality. 

This brings me to the work of Jeff Larrimore, Jacob Mortenson, and David Splinter. These government economists — one at the Federal Reserve Board and two from the congress’s Joint Committee on Taxation — have been putting out a series of papers which provide essential insights into how this general volatility of income interacted with the pandemic depression. I first learned of their work from a blog piece written by the economist JW Mason in 2021. In this work, they use Internal Revenue Service (IRS) data to track the incomes of individuals through time. Their January 2023 paper “Earnings Business Cycles: The Covid Recession, Recovery, and Policy Response” updates their results for 2021. I greatly hope they do another follow up paper with the 2022 and 2023 data. In my view, their work holds the key to understanding why the Biden administration’s narrative about the economy — at least when it comes to household wellbeing- was fundamentally misconceived. 

Consistent with the musical chairs analogy above, they report that “Over the last two decades, an average of 28 percent of workers had large [earnings] increases and 28 percent had large [earnings] declines each year”. “Large” in this case is defined as a 10 percent change from the previous year. It is very difficult to wrap our minds around such massive labor market volatility. In particular, it's challenging to take the time to truly understand the way fluctuations in aggregate unemployment rates and median wages — even calculated for just the bottom of the workforce- hides this massive labor market “churn”.

At this point readers might be thinking “if this is going on all the time, what is special about the experience since Covid started?” What is special is that the pandemic greatly impacted the volatility of labor market income and, at the same time, the expansions of the social safety net — particularly unemployment insurance — greatly reduced the volatility of overall income. This is especially the case for the bottom 50% of households, who always have far more volatile incomes than the top 50% of households. This is obvious when you think about it. Decently paying white collar jobs are overwhelmingly salaried, and much more stable than lower income jobs — which tend to pay workers hourly, and fire them easily.

As is well known, the Coronavirus Depression was unique for being services led. Simply put, the workplaces that needed to shutdown the most were retail outlets, restaurants etc. Economic downturns are generally led by other sectors. This uniqueness also explains why, in labor market terms, the Pandemic Depression was so extraordinarily regressive. Those who were in the bottom of the labor market in 2019 vastly disproportionately lost their jobs or lost labor hours — and thus income. As Larrimore et al. reports:

Between 2019 and 2021, 51 percent of workers who were in the bottom quintile had a large earnings decline. This is 7 percentage points above the 44 percent with large two-year earnings declines from 2017 to 2019. Among the top quintile, the 27 percent of workers with large earnings declines between 2019 and 2021 was 3 percentage points above the share with large earnings declines from 2017 to 2019. Similarly, large earnings increases among the top quintile were 2 percentage points less likely between 2019 and 2021 than in the two preceding years.

Recall that “quintile” is 20% of a group. You can also see how dramatically more unequal this downturn was compared to the Great Financial Crisis in their 2021 paper.

Compared to 2009, 2020 saw greater declines in labor market incomes among the roughly “bottom” 40% of workers and significantly smaller declines among the top 20%. Increases for the bottom 20% were roughly similar between the Great Financial Crisis and the Pandemic Depression: but beyond that point; the share of workers who had significant earnings increases was substantially greater in 2020; compared to 2009. Which brings us to the role of the pandemic programs shown in these dramatic charts from their 2023 paper.

In these charts we can see three very important things. First, the absolutely dramatic impact of the pandemic era programs meant that, when you combine them with labor market income, the share of workers experiencing large income declines between 2019 and 2021 was dramatically lower than over 2017 to 2019. In other words, better than a time period when the economy was seen as “good”. Meanwhile the impact on income increases was even more dramatic, with a significantly higher share of workers seeing large income increases; across almost the entire income spectrum. 

The key to these dramatic results is that both the “economic impact payments” and the unemployment insurance expansions were done in absolute dollar terms i.e. 1200, 600 & 1400 dollars for the three rounds of checks, Then an additional 600 dollars a week for unemployment insurance were reauthorized at a reduced amount of 300 dollars a week under Biden. This is generally why the impact across the income spectrum declines as you get to higher income groups — by definition 1200 dollars is a smaller percentage of household income as you move up the income ladder. However, clearly the role of additional household dependents in increasing higher income households payments from the direct checks and, in 2021, the child tax credit (CTC) played an important role in sustaining large income gains, until you reach the top 30% of households. 

The second apparent thing is that we can clearly see that the large economic support programs did help make the labor market far more equal between 2020 and 2021. 2020-2021 labor market outcomes are far more like 2018-2019 labor market outcomes than they are 2019-2020 labor market outcomes. Thus, there is a kernel of truth to the Biden narrative about the economy — back in 2021. It is important to note however, that a higher share of workers experienced large income declines in 2021 then in 2019, even if it's only a small difference, and very impressive considering that this happened after an extremely regressive recession caused by a pandemic.

Which brings us to the third apparent thing: these rosy narratives also look far different when you look at the results over a two year period. A much larger share of workers experienced large earnings declines between 2019 and 2021, then over 2017 to 2019. There is less, but still a significant gap for the workers experiencing large earnings increases between 2019 and 2021, then between 2017 to 2019. Doing better in the labor market than you did in 2020 is not saying all that much, and only partially covered the ground that was lost in 2020. The reason that households' financial wellbeing improved significantly between 2019 and 2021 despite such a dramatically regressive depression is purely because of the pandemic safety net.

This is all true and we haven’t even discussed the expiration of SNAP (food stamps) or the expiration of the medicaid expansion at the end of March 2023. Medicaid is an absolutely essential program that became even more essential in the pandemic. Hundreds of thousands of people have been disenrolled from Medicaid since March 2023 in crucial swing states alone, far larger than the thin gaps separating Trump and Harris in those states. Policy analyst Stephen Semler created a dramatic chart illustrating the continued wave of program expirations over the course of the Biden administration. It is astounding to look at.

The expiration of these numerous programs ripped away the insulation from the post-pandemic labor market, which Americans had just in time for energy, food and rent increases to significantly worsen the situation, especially for the bottom 50% of the labor market.

Which brings us back to price increases. The one weakness of Larrimore and his colleague’s work is that they take their rich microfounded data set and deflate (divide) them by an aggregate price index to produce allegedly “real” numbers. Those “large” increases and decreases of 10% are in “real” terms. This provides no additional information beyond their data, since we are interested in learning about individual households' economic experience beyond the aggregates- and CPI- or PCE- is an aggregate. As my colleague at Employ America Alex Williams presciently pointed out in a late 2021 report entitled “‘Real Wages’ and Aggregation: A Methodological Mess”- adjusting money incomes by an aggregate price index does not actually inform us about the consumption experiences or financial situation of individuals. 

In that respect, although money incomes without any reference to changes in money prices of important goods services would also be misleading, so-called “real values” do not actually provide greater information about individual economic circumstances. I would greatly prefer to have their data exercise recreated in purely monetary terms & have them show us how the results change for 5%, 10%, 15% & 20% income increases and declines. Analysts can make qualitative & multidimensional judgments of the impact of price increases in the absence of disaggregated price indices. Nevertheless, the fact that their data is tax return matched is a great improvement over other indicators and tells us roughly the right story even if I think a somewhat different approach would contain even greater insights.

Given that the price increases which are the most relevant for households budgets, especially the non-elderly among the bottom 50% of households, are food, energy and rent, it is likely that their results overstate the improvement in economic circumstances of lower income households, while overstating the decline in economic circumstances of the top 50%. When the role of program expirations like SNAP and Medicaid are additionally considered, the shift for lower income households is certainly even more dramatically negative.

These issues also provide us reason to doubt the great importance given to aggregate price indices when assessing the economic sentiments of the vast majority of households. One reason pundits have doubted the impact of “the economy” on the election results is that the growth of the Consumer Price Index slowed considerably from 2022. Besides how many households are disproportionately exposed to price increases in food, energy and rent — the perception of a rent as “high” rather than increasing is clearly far more important. Only a small percentage of people become new tenants each year and this fell precipitously in the pandemic. Thus the dramatic increases in rents in 2021 and 2022 for new tenants are a continuing shock to people renewing leases, or having to move. According to a Federal Reserve Board survey (which I will return to), 27% of renters who moved between roughly the beginning of November 2022 and the beginning of November 2023 moved because of rent increases at their previous home.

It’s also important to recall that ordinary households do not respect the boundaries of the Bureau of Labor Statistics methodologies. Most Americans are not aware, nor do they care, that mortgage interest rates were removed from price indices by the Bureau of Labor Statistics in 1983. They also do not care that credit card interest rates are not in these indices either. As far as they are concerned interest rates are a price they pay — and both those interest rates went way up over the past few years. 

It’s also important to remember that CPI and PCE, by their very nature, are weighted to the consumption expenditures of higher income households. That’s because they spend more. The differences between households with a below current market interest rate mortgage, those with market interest rate mortgages, those with no mortgage and renters can be dramatic. The inclusion of “owners equivalent rent” distorts things further, by imputing such large flows of incomes to homeowners and downplaying the good financial circumstances of so many homeowners — especially those with fully paid off homes. Recall that “owner’s equivalent rent” is a construction where government accountants calculate what the “market rent” of owner-occupied housing would be, and assume that homeowners are charging themselves market rents — which they are of course not doing in reality. This income is attributed to them in GDP, and is “weighted” along with actual household expenditures in aggregate price indices.

In other words, homeowner’s incomes are represented as far higher than they are but all of this “income” goes right into paying “rent” on the “expensive” homes they themselves own- expensive, that is, because it is valuable. Homeowners truly are rapacious landlords to themselves —at least in the offices of the Bureau of Labor Statistics (for now). Don’t misunderstand me — many government economists are aware of the conceptual issues and distortions this accounting gimmick creates. For example, a very good Bureau of Labor Statistics paper from last year provides insight into these problems by exploring the construction of “Housing Cost Indexes” in the United States.

The main distinctive feature of these indices, besides going back to elements of the pre-1983 approach to housing costs, is that they give every household’s expenditures an equal weight — rather than giving households who spend more greater weights. In their sample 41.4% of households are homeowners with a mortgage, 29.1 are homeowners without a mortgage and 29.6 are renters (this seems to overestimate homeowners a little). Price indices for each of these groups' consumption baskets would clearly diverge significantly (especially since homeowners with paid off mortgages are overwhelmingly elderly). 

The "relative importance” of different expenditures in their approach, disaggregated in this way, already points to how significantly divergent their spending patterns are — and thus the impact of price increases of various different products. Relative importance is simply an average of what percentage that group of people spend on x or y good or service. However, we don’t actually have price indices to go along with these relative importance “weights”. And the more you disaggregate with current data, the larger the errors become. Data collection methods would have to change to overcome these problems to produce high quality disaggregated price indices that inform us about the outcomes of specific groups. In other words, we would have to oversample a number of subgroups in order to generate accurate data fit for the purpose of giving better insights about the experiences of individual households.

Nevertheless, from the evidence we’ve examined so far we can already be fairly certain of a number of things. 

1. Income volatility, especially downwards income volatility, greatly increased when the pandemic era programs expired. 
2. This was worsened by price increases, especially energy, food and rent increases in 2022 and the rent increases, because of the structure of the rental housing market, continued to impose new economic pain on a growing percentage of renter households even as rents for new tenants stabilized
3. Households treat interest rates as a price, and thus to them price increases were even more dramatic in 2022 than the CPI increases. Additionally, similarly to rent, higher interest rates impose continued economic pain on a growing percentage of households as the interest payments remain high, and grow when borrowers need to refinance or new borrowers enter these markets
4. Households, particularly lower income households, experienced these salient price increases and high prices, the running through of financial assets accumulated in 2020 and 2021 & the loss of the safety net expansions as a worsening economic situation, regardless of what the headline numbers said.

Which brings me to the final point of this piece. It is the combination of all these different economic factors which make me very confident in surveys where households report their own economic wellbeing. No, households do not have a very good understanding of aggregate economic indicators, and are wrong to think we are in a recession. On the other hand, as we have seen, the pundits do not have a very good understanding of how aggregate economic indicators relate to individual economic circumstances, so lets call this even. The evidence also seems to suggest that issues like partisanship or media ecosystems mostly impact the perception of what’s happening to others, or in the local or national economy — rather than individuals' perception of their own financial circumstances.

Enter the Federal Reserve Board’s Survey of Household Economics and Decisionmaking (SHED). Each May the Fed puts out a report based on this survey entitled “Economic Well-Being of U.S. Households”. From my point of view, the key question from this survey is “Compared to 12 months ago, would you say that you (and your family) are better off, the same, or worse off financially?” Given these various factors, especially the Larrimore et al. paper, I would expect the answers to this question to be surprisingly muted in 2020, and maybe even outright good in 2021, with a dramatic reversal in 2022 and 2023. And that’s exactly what we find when we look at this data.

As you can see above, the percentage of people who said they were better off than 12 months ago dropped significantly from 2020-2021 to 2022-2023. Most dramatic, however, is the drop off for the percentage of people who said they were worse off. Fourteen additional percent of adults said they were worse off than 12 months ago between 2021 and 2022. This number did not decline significantly in 2023, either. If you treat these statistics as similar to the net favorability polling applied to politicians, we can produce a statistic called a “net better off rating” — the percentage of people who say they are better off financially than a year ago minus the percent who say they are worse off. As you can see above, the “net better off rating” of overall household financial wellbeing dropped a staggering 21 points in 2022. A net better off score of negative 15 points, even negative 11 points, is very bad. Those numbers make running on a “good” economy a catastrophic move. 

Nor does this seem to be the result of rosy imaginations about the economy under Trump. For one thing, the Larrimore et al. papers make clear that the economy during the first three years of the Trump administration genuinely did have much better labor market outcomes than the economy of the 2020s. That was true in ways hidden by the reliance on statistics like median real wages and headline unemployment rates:

However, one-year improvements include mean reversion of prior-year losses, which is why we also consider two-year changes. Over the two-year period from 2019–2021, large earnings increases before fiscal relief were 1 percentage point less common than large earnings declines (34 percent vs. 35 percent). For comparison, in the pre-Covid expansion years from 2017 to 2019, large increases were 7 percentage points more frequent than large decreases.

Whether Trump was responsible for these labor market outcomes, or they reflect a more complete recovery from the Great Financial Crisis after many agonizing years does not change the fact that people are not simply inventing that they had better personal economic circumstances pre-pandemic, once the safety net expansions faded.

One key sign from these surveys that partisanship or rose-tinted glasses about the prepandemic world are not driving the Federal Reserve survey’s results is that in 2021 the survey asked the same question, except for the change from two years ago rather than one. The result was 36% said they were better off than 2019, and 24% said they were worse off. This comes to a “net better off rating” of +12 which is extraordinarily impressive, when you consider the economic dislocations of Covid. By greatly underestimating the extraordinary power these programs had, pundits and politicians alike have greatly underestimated the effects of their expiration. This meant they didn’t predict the wave of discontentment that would inevitably emerge — and has now emerged.

You can argue that the safety net expansions, especially after CARES act, were primarily a result of Democrats in the face of Republican opposition. But this is of little relevance if voters are not hearing this narrative, or being told that you are fighting to bring back their economic support. Instead they heard a bevy of information about foreign policy amidst foreign policy crises — and nothing about bringing back these programs. 

Instead, once Russia invaded Ukraine, Biden dropped all mention of safety net programs, and started touting his failure to get Build Back Better passed as a success. That means he touted deficit reduction that only happened because of that political failure. Again, policy analyst Stephen Semler has produced another incredible chart of Biden’s rhetorical pivot after the invasion of Ukraine, as expressed through his various twitter accounts. In the absence of a narrative about these economic support programs expiration, voters assume that “someone else” is getting their money —whether it's overseas military spending or “immigrants”. 

All of this is true without even approaching the central question that hung over Biden his entire presidency: his age. If the United States had one consensus during the Biden years, it was that Biden was too old to run for reelection. The last minute switchover to Harris was likely too short to dramatically change messages, or gain credibility as a “change” candidate in a “change” election In fact, it’s that Harris didn’t even really try. This is understandable given the unusual circumstances, but it was electoral poison given that there is nothing like the rage of falling financially behind while being told the “economy” is going great. The Democratic party failed to take advantage of the uniqueness of the U.S. political system. That is, the separation of the legislative and executive branch makes it possible to credibly run as a change candidate, even if you are a member of the same political party as the president. This isn’t true for the rest of the incumbents who lost elections around the world in 2024. 

The good news is, if aggregate economic indicators can be so disconnected from household financial wellbeing, then great expansions of the safety net do not require big increases in overall demand- perhaps even at all. Even 600 dollars a week added back to unemployment insurance would mean far less spending with unemployment rates so low. Meanwhile programs that squarely target poverty are not very expensive. That is, after all, the core meaning of the phrase “social insurance”- to create programs that make catastrophic income declines far less likely. Someone should get on that.