Policy Hub: Macroblog provides concise commentary and analysis on economic topics including monetary policy, macroeconomic developments, inflation, labor economics, and financial issues for a broad audience.
Comments are moderated and will not appear until the moderator has approved them.
Please submit appropriate comments. Inappropriate comments include content that is abusive, harassing, or threatening; obscene, vulgar, or profane; an attack of a personal nature; or overtly political.
In addition, no off-topic remarks or spam is permitted.
August 20, 2021
How Does a Household's Exposure to Monetary Policy Vary over the Life Cycle?
A recent study by Feiveson et al. establishes the Federal Open Market Committee's interest in the distributional effects of monetary policy. The size and the composition of household income exhibit large variation over the life cycle, so it is likely that household exposures to monetary policy also depend on age. This post summarizes new research by Daisuke Ikeda and me that uses a life cycle model to measure the age profile of household exposures to monetary policy. In the model, a higher nominal interest rate increases the wealth and consumption of households between the ages of 60 and 80, but it reduces the wealth and consumption of younger working-age households and the oldest retirees. The former group also has the highest net worth, and it follows that net worth and consumption inequality increase in the model.
Our new research took as its jumping-off point the premise that a household's age affects its economic opportunities. Both the size and the sources of income vary with age. On average, 55-year-old workers have higher earnings than 25-year-old workers and also higher earnings than 65-year-old workers. Other research by us documents this result for the United States and Japan, but age-earnings profiles are hump-shaped in other high-income economies, too. Beyond age 65, an increasing fraction of individuals in high-income economies have low or no labor earnings as they transition into retirement. Retirees have no labor income, and an important source of income for them is their public pension, which typically only replaces a fraction of their previous labor earnings.
Individuals understand these constraints and cope with them by making asset-allocation decisions. Table 1 depicts the age profile of household net worth and a decomposition of net worth into two categories: liquid assets and illiquid assets. Liquid assets include deposit accounts, CDs, bonds, and all loans, while illiquid assets consist of physical assets like homes, cars, and financial assets such as stocks, which are more costly to acquire and sell. We use Japanese survey data because they provide considerable detail on the various components of household net worth. Younger households have low net worth and negative holdings of liquid assets (they are net borrowers), but they hold positive amounts of illiquid assets. Net worth increases with age up to retirement, which typically occurs between the ages of 60 and 69 and then declines during retirement. Older working-age households and younger retirees hold positive amounts of both liquid and illiquid assets. A limitation of our data is that they don't provide details about asset holdings of the oldest households. Indirect evidence we discuss in Braun and Ikeda (2021) suggests that some older households have negative holdings of liquid assets too.
Note: The age of a household is indexed by the age of the household head. Liquid assets are net of all household borrowing, and net worth is the sum of liquid and illiquid assets. All numbers are divided by income of the 50–59 age group.
We expect that similar patterns also occur in other countries. However, the specific magnitudes of the age profiles of income, net worth, and their components will depend on institutions in a given country. For instance, households in countries that offer free tuition for higher education will have less student loan debt.
A tighter monetary policy (in other words, a higher policy nominal interest rate) is generally associated with higher real interest rates on deposits and loans (liquid assets), weaker performance of stock and real estate markets (illiquid assets), and slower growth in employment and wages. Given that the size and composition of income and net worth vary with age, one might surmise that a household's overall exposure to monetary policy also depends on its age. Retired households, for instance, may gain because they have no direct exposure to the labor market and hold large positive amounts of deposits whose return goes up. Young working-age households, in contrast, may lose because they have low net worth, loans, and low labor earnings.
Unfortunately, finding data that can be used to directly assess these hypotheses is challenging. In the United States, there is reasonably good survey data about how labor income and financial assets vary by age, but much less information about the size and value of household holdings of physical assets like homes, cars, and TV sets. Moreover, even in countries like Japan, where reasonably comprehensive survey data are available, a cross-sectional snapshot is produced only once every five years. Even if we can identify exogenous changes in monetary policy, we lack high-frequency data to measure how household exposures to monetary policy vary by age.
An alternative approach is to use an economic model. To see how this works, we define wealth as a household's net expected present value of future income from labor, assets, and the government. Wealth is an important economic concept because standard economic theory predicts that a household that sees its wealth increase from a tighter monetary policy will consume at least a fraction of its bonus. Conversely, a household whose wealth falls will consume less. Recent work by Auclert builds on this insight. He uses a model to specify the dynamics of household income and decompose a household's consumption response to a change in monetary policy into four components:
- The income component captures the impact of changes in monetary policy on labor and government income.
- The unexpected inflation component captures net capital gains or losses associated with holdings of nominal assets. For instance, most government debt is nominally denominated, and a change in monetary policy affects the inflation rate and thus the real value of this nominal asset.
- The unhedged real interest rate component captures net real capital gains on household assets that are coming due at the time of the shock. For instance, a higher real return on deposits is good for a saver who has no loans, but a higher real interest rate can be bad for a borrower who enters the period with a maturing loan and faces a higher real cost of paying it off.
- Finally, the substitution component captures how a change in the interest rate affects a household's tradeoff between consuming today and saving today, which allows it to consume more tomorrow.
In our working paper, we propose a model designed to measure how household exposures to monetary policy vary over the life cycle. We specify the model to reproduce the main features about how household income, net worth, and portfolio allocations vary over the life cycle using data from Japan. Our model is rich in the sense that households are active for up to 100 years. They work and make asset-allocation decisions over time and interact in markets with households who have different ages and thus different asset-allocation priorities. Further, we model a government that taxes households, issues nominal debt, and runs a public pension program. Finally, the monetary authority sets the nominal interest rate on liquid assets using a simple rule. Fortunately, the model also has sensible implications for how nominal and real interest rates, wages, and government income respond to a tighter monetary policy.
Our model may sound rather sophisticated, but we make many simplifying assumptions. For instance, we are silent about what determines cross-sectional differences in income and wealth among households with the same age. In addition, households have only two assets that they can use to borrow or save. These simplifications make it easier to understand how age affects a household's exposure to monetary policy.
Figure 1 reports the age profile of household consumption responses to a surprise tightening in monetary policy in the year that monetary policy is tightened (left panel) and its decomposition into the four components we discussed above (right panel).
Source: Braun and Ikeda (2021)
The sign of the consumption response varies with age in figure 1. Households close to age 68 are increasing their consumption in response to higher wealth, while older retirees and younger working-age households are facing wealth losses. Another way to ascertain differences in exposure is to measure the magnitude of the consumption response. Households around age 30 reduce their consumption most, households close to the retirement age of 68 in the model increase their consumption most, and households that survive to about age 100 reduce their consumption. The magnitude of the consumption response is an imperfect measure of exposure because net worth also varies by age, as reported in table 1. In the model, the two groups who are reducing their consumption most have relatively low net worth. Younger workers are borrowers, and old retirees of age 100 have lived well beyond their expected life span and exhausted their savings. Thus, the biggest negative exposures to a tighter monetary policy in the model are among younger workers and oldest retirees.
The right panel of figure 1 reports the Auclert decomposition of consumption responses. For younger working-age households, the negative income component and the negative intertemporal substitution component are the two biggest factors. They have lower labor income and are at the age of their life cycle where they are accumulating assets, so movements in interest rates are particularly important for them. The other two factors are less important because their net worth is low. For households between 60 and 80, the income component is small, and the two asset-income components primarily drive their consumption responses. A lower inflation rate benefits this group because they are holding relatively large positive positions in nominally denominated liquid securities to provide for their retirement. The unhedged real interest rate component (unhedged R in the chart) is large because these households are savers and are at the stage of the life cycle where they draw down their assets to smooth their consumption during retirement.
In the model, life expectancy is 83 years, and households who survive beyond this age experience declines in all four components. They have been consuming their savings since age 68 and have low net worth. Also, some members of this group have debt. This age group also receives lower net income from the government. Government labor tax revenue is down and interest rate expenses on government debt are now higher so net government transfers to households fall, and this decline is significant for the oldest households in the model.
Taken together, these findings imply that inequality in net worth increases in the model in the year that monetary policymakers tighten policy. The highest net-worth age groups see their net worth increase, and the age groups with the lowest net worth see it fall. Consumption inequality also increases in the model because households with lower net worth tend to adjust their consumption by more than households with high net worth.
Hopefully, our findings have piqued your interest and left you with new questions. How large and persistent are the changes in inequality? What are the properties of an easier monetary policy? Does the amount of government debt in the economy matter? What about the effective lower bound on the nominal interest rate? I encourage you to read our working paper to find out.
I conclude with an old saying from economics: for each borrower, there is a lender. In our model, monetary policy alters interest rates, and a higher interest rate affects borrowers and lenders differently. It's a burden on younger working-age households and on the oldest retirees who are borrowers, but it's a boon for households close to age 68 who are the savers who provide the funding for the loans to the other two groups.
August 10, 2021
Do Rising Retirements during COVID Reflect Demographic Trends?
Data from the Current Population Survey tell us that, in the second quarter of 2019, 47.8 percent of those aged 55 and older said they didn't want a job because they were retired. By the second quarter of 2021, that share had risen more than 2 percentage points, to 49.9 percent, which is an increase of around 2 million retirees over what would have been expected if the retirement rate for those aged 55 and older had not changed.
These data raise the question of how much of the increase in retirements is over and above what would have been expected based on the ongoing aging of the baby boomer generation—the movement of more people into ages that are more likely to retire. In other words, did the COVID-19 pandemic contribute to an increase in retirements?
The Atlanta Fed's Labor Force Participation Dynamics tool, which we recently updated with data through the second quarter of 2021, allows us to investigate the source of the change in retirement. The increase in the overall retirement rate for those aged 55 and older can be broken into two parts. The first one is the part due to a shift in the distribution of age, sex, race/ethnicity, and educational attainment toward demographics with higher retirement rates. For example, a 65-year-old is more likely to retire than a 63-year-old, and we have more 65-year-old people today than two years ago. The second part is the increase due to higher retirement rates within the age, sex, race/ethnicity, and educational attainment groups.
To illustrate how the decomposition works, let's look at just two factors: age and sex. The following table shows the average retirement rates of men and women aged 55 and older by five-year age groups for the second quarters of 2019 and 2021. The numbers in parentheses show the share of the 55-and-older population in each age/gender group. For example, in the second quarter of 2019, 51.5 percent of women 55 and older were retired, and women made up 53.7 percent of the overall population of people 55 and older.
Looking down the columns of the table, notice that for both men and women, retirement rates are much higher for those in their 70s than in their 60s—and much higher for those in their 60s than in their 50s. This matters because, comparing 2021 with 2019, the share of the population in the older of the age groups for both men and women has increased. This fact alone puts upward pressure on the overall retirement rate for the 55-and-older population between 2019 and 2021.
But in addition to an aging 55-and-older population, the table above shows that retirement rates have also increased within the age/gender groups. Looking across the age rows of the table we see that the retirement rate for each age/gender group is higher in 2021 than in 2019. So not only are there more women and men of ages that have higher retirement rates, the retirement rates themselves have increased.
Chart 1 displays the results of the complete decomposition. The blue line is year-over-year change in the retirement rate of those 55 and older going back to the second quarter of 2006. The orange bars represent the part of the change in the overall retirement rate accounted for by changes in the demographic composition (the distribution of age, sex, race/ethnicity, and educational attainment), while the green bars depict the contribution to the overall change from changes in retirement rates within the demographic groups (labeled as behavior).
Notice that up until 2020, behavioral changes were generally contributing to lowering the overall retirement rate of the 55-and-older population. The loss of retirement savings during the Great Recession was arguably an important factor in reducing the ability to retire during that period. At the same time, demographics were also putting mild downward pressure on retirement, with the leading edge of the baby boomer generation still within an age range with relatively low retirement rates. However, since 2013 underlying demographic shifts have been putting upward pressure on the overall retirement rate.
During the COVID-19 pandemic, demographic and behavioral factors appear to have contributed roughly equally to the rise in retirements. Perhaps, for some baby boomers who were already likely to retire within a few years, the pandemic created an incentive to retire sooner than they might have otherwise. A look at the Federal Reserve's Distributional Financial Accounts Overview shows that the annual growth in the net worth of those 55 and older now puts them, on average, in a much better financial position to retire than was the case during the Great Recession (see chart 2).
The ongoing aging of the baby boomer generation will continue to put upward pressure on the retirement rate over the next few years. How much the recent behavioral change will persist is much less clear, and a great deal will undoubtedly depend on the future path of the pandemic and the financial resources of older Americans. The Atlanta Fed's Labor Force Participation Dynamics tool will allow you to investigate the changes for yourself—with data for the third quarter of 2021 available sometime in October—but I'll be back to discuss my own findings with you here.
July 15, 2021
Onboarding Remote Workers: A Hassle? Maybe. A Barrier? No.
As the need for social distancing recedes and government restrictions ease, most people look to regain some notion of normalcy in their day-to-day lives. At the same time, many workers have come to like their office away from the office. And the COVID work-from-home experiment has gone well enough for it to stick around, possibly in a hybrid form. Those are key conclusions in a recent study (by three of this post's authors) titled "Why Working from Home Will Stick."
One frequent concern we hear about remote work is the challenge of hiring and onboarding new employees who rarely—or even never—set foot on the employer's worksite. Yet our evidence suggests that these challenges are modest and aren't a big barrier to finding, onboarding, and integrating new employees.
During the past two months, we asked executives participating in our Survey of Business Uncertainty (SBU) about their experiences hiring remote workers and integrating them into their organizations (see the three charts below). The share of new hires who work almost entirely from home (rarely or never stepping onto business premises) was 15.8 percent, a figure nearly identical to the share of all paid workdays performed at home in earlier pandemic-era snapshots from the SBU. Moreover, the share of new hires varies across industries, just as we'd expect based on research by Jonathan Dingel and Brent Neiman that flags which jobs can be performed remotely. The Survey of Working Arrangements and Attitudes (whose data underpin our aforementioned recent study, "Why Working from Home Will Stick") also suggests that new hires are at least as likely to work from home as the general population.
So as we see, new hires are working remotely to the same extent, and in the same proportion across industries, as incumbent staff. These findings suggest that integrating and training new remote workers isn't a big barrier to hiring. But it must be a pain, right? Otherwise, why all the fuss?
And here, the unanimity dissipates. Of firms with new remote workers, 60 percent say the integration process is more challenging. On average, SBU respondents say it takes about a month or so longer to fully integrate remote-only employees, though the spread about that average is pretty wide—ranging from six weeks less to six months longer.
If you're at a firm that finds it challenging to onboard remote employees, perhaps you'll find some solace in the fact that most of your new staff appear not to notice. In the The Survey of Working Arrangements and Attitudes, 42 percent of workers hired into fully remote positions during the pandemic say that adapting to their new jobs has been neither easier nor harder than adapting to in-office jobs before the pandemic. The other 58 percent are distributed similarly over "easier" and "harder" (see the chart).
Stepping back and taking a broader look, many observers wonder why aggregate U.S. employment remains 6.8 million below its prepandemic peak, despite record numbers of job openings. On the list of potential reasons for this shortfall—such as lingering concerns about the virus, generous unemployment benefits, inadequate childcare options, and more—it appears you can cross off the difficulty of onboarding and integrating remote workers.
July 14, 2021
Are Labor Shortages Slowing the Recovery? A View from the CFO Survey
The economic recovery from the pandemic-induced downturn of 2020 has been swift in terms of overall spending. It took about one year for real gross domestic product to return to its pre-COVID level. Firms in the latest CFO Survey expect the pattern of strong sales to continue for 2021 and 2022. At the same time, firms report—by a large margin—that their top concern is a shortage of available labor. In this post, we investigate the extent to which labor-availability problems are restraining sales revenue growth. Said another way, could the recovery in spending be even stronger if labor shortages could be resolved?
In this quarter's CFO Survey, we asked firms' financial executives a series of questions designed to uncover just how much a labor shortage has crimped their revenues. We estimate that the labor shortages have reduced economywide sales revenue by 2.1 percent, which suggests that the lack of available labor has reduced nominal private-sector gross output by roughly $738 billion, on an annualized basis (or $184 billion per quarter; see the note below).
Difficulty finding new employees for open positions is widespread, as indicated by three-quarters of the respondents to the July 2021 CFO Survey (see chart 1), a finding consistent with the record levels of unfilled job openings in the Job Openings and Labor Turnover Survey from the U.S. Bureau of Labor Statistics.
Among firms that struggled to find workers (nearly 40 percent of our overall panel), slightly more than half reported that their inability to find employees cost their firm revenues. It also appears, as chart 2 indicates, that smaller firms have been disproportionately affected by labor shortages. Of the small firms (fewer than 500 employees) that struggled to hire employees, nearly 60 percent indicated that labor shortages caused their revenue to suffer, compared to 40 percent of large firms.
Digging even deeper, we posed the following question to the firms indicating that their inability to find employees has reduced revenue: "By roughly what percentage would you say your firm's revenue has been reduced due solely to your inability to find new employees?" The response was a fairly consistent 10 percent across industry groupings for this subset of panelists Aggregated across the full panel (thereby also accounting for firms that have not been affected by hiring difficulties or revenue effects), this decline indicates that labor shortages cost the economy 2.1 percent in nominal sales revenue (or private-sector gross output). Nationally, this impact translates into an annual reduction in gross output of $738 billion.
There does not seem to be just one reason for why workforce participation has remained lower than it was before COVID despite the record number of job openings. For instance, households have had ongoing virus concerns, and many have struggled to find childcare. Also, the share of older workers choosing to retire has risen, and government support payments have made not working relatively less costly than in the past. Nonetheless, the CFO Survey results suggest that labor shortages have burdened firms significantly and could further restrain aggregate economic growth if not resolved.
Note: The figure given in this post earlier was erroneous. The correct estimate of the reduction is $738 billion (or $184 billion per quarter). Nominal gross output for all private industries was $35.18 trillion in the first quarter of 2021 (at seasonally adjusted annualized rates). Multiplying that by 0.021 results in $738 billion (and dividing that by four yields $184 billion per quarter). The U.S. Bureau of Economic Analysis defines nominal gross output for private industries this way: "Principally, a measure of an industry's sales or receipts. These statistics capture an industry's sales to consumers and other final users (found in GDP), as well as sales to other industries (intermediate inputs not counted in GDP). They reflect the full value of the supply chain by including the business-to-business spending necessary to produce goods and services and deliver them to final consumers."
- Business Cycles
- Business Inflation Expectations
- Capital and Investment
- Capital Markets
- Data Releases
- Economic conditions
- Economic Growth and Development
- Exchange Rates and the Dollar
- Fed Funds Futures
- Federal Debt and Deficits
- Federal Reserve and Monetary Policy
- Financial System
- Fiscal Policy
- Health Care
- Inflation Expectations
- Interest Rates
- Labor Markets
- Latin AmericaSouth America
- Monetary Policy
- Money Markets
- Real Estate
- Saving Capital and Investment
- Small Business
- Social Security
- This That and the Other
- Trade Deficit
- Wage Growth