Author: Dr. Michael Dooley | December 10, 2020
Further stimulus has come to be seen as necessary for a continuing economic recovery, but what if it isn’t? The implications of a service sector-specific shock are consistent with what we have observed: continued recovery in the face of fading fiscal stimulus.
The ability to spend income on services has been significantly reduced (think restaurants that have closed), and that income will either be spent on other goods (helping the recovery) or on assets (pushing prices higher). We think asset prices and perhaps aggregate demand can continue to rise even without another large stimulus package.
To be sure, there are over 13 million Americans currently receiving supplemental income through programs authorized during the initial lockdowns. These programs are set to expire at the end of the year. While our outlook for the economy is benign even if these programs run out, the impact on individuals receiving benefits would be anything but benign. We think the government should, can, and will renew those benefits come 2021 but the economy, as a whole may not need another large fiscal intervention.
The US government responded quickly and in considerable scale to support incomes of households and firms affected by the Covid-19 pandemic. A striking feature of the government support is that it was heavily frontloaded in April and May of 2020 and has since been greatly reduced. (See Figure 1). A widely held view is that failure to extend support programs at the federal level risks a recession in 2021. With the resurgence of the Covid-19 pandemic in November and the political chaos associated with the presidential transition, evaluation of the consequences of federal transfer payments has become the key ingredient in the outlook for the US and the world economy.
Figure 1: Progression of Government Transfer Programs
The obvious potential impact on the economy as a whole is a decline in the service sector’s demand for output from the rest of the economy. Without government programs, it seems quite likely that service sector businesses (hair salons, restaurants, etc.) would be unable to stay in business, thus depriving their employees of income and the ability to purchase goods. The initial federal assistance programs may have offset this potential demand shock to the economy as a whole by maintaining the income of service sector households and firms.
But, if this were the whole story, we should be seeing the demand shock by now as the assistance program has melted away and expectations for renewal of the assistance have become less certain. As shown in Figure 2, goods sector output has continued to shoot higher even as services output (the majority of the economy) remains well below where it was in 2018. Using data from 1945 to 2020, the >40% fall (at an annualized rate) of services output in the second quarter was a 21 standard deviation event and was a significantly steeper drop off than seen in the Great Depression.
Figure 2: Service Sector Shock
How is it possible that the services sector can see its output fall substantially, but goods output increase, and aggregate output remain remarkedly stable? We think the answer is that a sector supply shock also delivers a powerful positive shock to aggregate demand as the economy adjusts to increased demand for financial assets by goods sector households, thus pushing down required rates of return. This feature of a sector supply shock might account for the recent strength in equity and housing markets.1
As a starting point, let us assume that Covid-19 can be characterized as an exogenous shock to output to the service sector of the economy. The households employed in the service sector remain as consumers and are active in credit markets, but their output cannot be produced or sold. It follows that, for some interval, income to capital and labor in the service sector is zero and the consumption of households that own the capital and labor depends on sales of financial assets (net borrowing) or transfers. Consider the following simplified example: due to an exogenous shock all restaurants are closed. Consumers can’t buy meals at restaurants, the staff has to stay home without a paycheck, and the owners (capital) do not see any income. However, both the staff and owners are still active consumers (they may buy food at the grocery store) and they will have to use cash in their savings account (selling a financial asset) to make purchases.
Output and income in the goods sector is not immediately affected by the shock. The only way the goods sector households will directly notice the shock is that they cannot buy the service sector’s output.
Let us assume that the goods sector’s income and output is predetermined but expenditure on services is now zero so they have to reallocate income towards goods and/or financial assets.2 In other words, I want to use part of my paycheck to go to a restaurant, but it’s closed. I can either buy groceries with it (goods) or put it in the bank/stock market (financial assets). For now, assume no government so net asset sales between the two sectors must be zero.
Following the shock, the service sector households’ income is zero. In order to consume goods, they must sell assets to households in the goods sector. If the restaurant can’t open then the server can’t earn money. If the server needs to buy something they must: a) sell a stock, b) borrow money, or c) use cash in their bank account. All of these options amount to selling assets to the goods sector.
The demand for services goes to zero. The supply of goods is initially unchanged but demand for goods can go either way. The direction of demand for goods depends on two quite different types of behavior.
The more obvious is the preferences for goods relative to services in all households. The second is the preference for assets relative to consumption of goods.
Following the shock, households in the goods sector must shift consumption away from services toward goods and/or financial assets. Assume for now that their preference is to make a small shift toward goods and a large shift toward assets. In national income accounting terms, their demand for goods would rise a little and their desired saving goes up almost as much as their former consumption of services.
Service households are clearly motivated to offer to sell assets and buy goods since the alternative is no consumption at all. They might for example want to sell enough assets to consume about the same level of goods as they did before the shock.
With these special preferences the financial asset markets might clear with minor changes in interest rates and asset prices. Households in the goods sector might "succeed" in saving most of the income they cannot spend on services. If the asset market clears, service sector households necessarily dissave by exactly the same amount and in turn "succeed" in consuming goods in an amount a little less than their former income.
But there is no reason to believe that the value of assets service sector households want to sell would be equal to the value of assets goods sector households want to buy at pre-shock interest rates and asset prices.
Suppose service sector households have no net financial assets when the shock hits, the shock is expected to be long lasting, and that the ability of service sector labor and capital to move to the goods sector is limited. In this, perhaps realistic case, prices in the credit markets will have to adjust. If goods sector households want to save (buy assets) most of their former service expenditures then credit constrained service households may only be able to offer credible assets equal to say half of what the goods households want to buy.
In the asset market there is now excess demand. One market response might be a decline in the risk free interest rate but with rates already near zero this is an unlikely outcome. A more likely alternative is bidding up asset prices more broadly, thereby reducing risk premia. If we assume all the resulting capital gains are saved (not a bad assumption) the goods sector households will succeed in saving about half of what they formerly spent on services. This saving is matched by dissaving in the service sector. Aggregate savings rates in the GDP accounts will not be affected if capital gains are included in income. We should expect a rally in equities and rising prices for existing homes and other durable goods.
In the goods market anything is possible. Demand for goods by service sector households is now half of its former level. This is the aggregate demand shock that we and others have expected in cases where borrowing by households is constrained. But this may be more than offset by reactions of goods sector households to the boom in asset market prices.
The demand for goods that provide future consumption will rise as interest rates fall and asset prices rise. Demand for durables (cars, home remodels) could rise a lot as goods sector households react to lower interest rates and rising asset prices. Moreover, as the constraint on service sector borrowing increases and generates reduced demand for goods, so does the excess supply in the asset market and the boom in asset prices.
The only clear outcome is that service sector households are much worse off if they are credit constrained. In the next section we turn to the role of the government in dealing with the shock. We conclude that the government should assist the affected households and firms but should also be concerned about the perhaps temporary boom in asset prices.
This framework is useful because it is easy to add the government. When the government takes a loan, that is, sells an asset (runs a debt financed fiscal deficit) it has supplied an asset to the asset market. The likely buyer of that asset in the sector shock scenario is the goods sector. If the Treasury then transfers the proceeds of the sale to the service sector there is no further change in the asset market but the service sector households can now buy goods. This transaction is equivalent to the service sector household taking a loan, except that repayment comes from future taxes instead of future income. The US government has offset credit constraints faced by service sector households by selling assets and transferring the proceeds to the service sector.
As shown in Figure 3, government intervention in the credit markets has been an important feature of the current crisis. The government can supply assets to goods sector households and transfer the proceeds to the service sector households. The government in effect has replaced the private financial market.
Figure 3: The Government Steps In
Source: SIFMA, Bloomberg
But should the government play this role? To date, the apparent objective for Covid-19 relief is to support consumption by service sector workers. If these workers are credit constrained the above analysis suggests that the intervention increases the demand for goods by service households.
But the government’s bond sales also reduce excess demand in the asset markets and reduce the boom in asset prices. Transfer policy created considerable supply in the asset markets April – September. Sales of securities to goods households satisfied their desire to buy assets. This limited the boom in asset prices and fall in interest rates.
After September, Covid-19 assistance declined and excess demand in the asset market increased. If there is no additional government stimulus it is very likely that demand for goods by service households will fall. But credit markets could swing further toward excess demand as goods households look for assets to buy instead of services.
The net result for aggregate demand is not clear. The continued erosion of government transfers might drive a very good period for asset prices, low interest rates, and a more intense boom period for housing and durable goods. If this outcome comes to pass, there are important questions to ask about financial stability when the shock subsides. Stay tuned.
In a more familiar economy wide aggregate demand shock, the counterpart to a negative demand shock is an unexpected "investment" in inventories. While Keynes noted the expansionary effect of lower interest rates (caused by consumer demand for assets), he expected the effect to be overwhelmed by the contractionary effect of inventory accumulation by firms, thus leading to the traditional propagation of an initial shock into lower output, higher unemployment, and a recession. Today, services sector output has been curtailed and there is no inventory accumulation.
For convenience we measure all financial market transactions as sales or purchases of assets. Borrowing, for example, is the sale of a financial asset by borrowers and lending is the purchase of a financial asset. When service households cash out their home equity, they take a loan collateralized by their real estate. They have "supplied an asset" to the market. When someone in the goods sector buys the asset, perhaps after it is sliced and diced, the goods sector demands that asset. This is a useful way to think about it because any set of financial transactions undertaken by a sector over some time interval, including capital gains and losses, can be summarized by the net addition to or subtraction from the sector's net asset position.