The other law of large numbers

When it comes to recessions, your large response is never large enough

TLDR:
- The headline numbers on the Covid19 fiscal responses are large
- But we’ve never seen a hit to consumption like this
- A chunk of it won’t actually increase incomes
- Tax and Saving mean a lot of the income won’t be spent
- If spending isn’t maintained, a wave of bankruptcies is coming

All expenditure is someone else’s income, but the reverse is not true. Income is not necessarily spent. This is at the heart of why recessions are so hard to fight - and why government interventions are rarely big enough.

US personal consumption expenditures were $15 trillion in 2019. The government on the other hand consumed about $4 trillion. $3.5 trillion went into forming new capital. If we start to put some plausible numbers on how these things have declined in the covid19 crisis the scale of the problem is immediately apparent. Investment post GFC fell 28%:

White: Gross Private Domestic Investment. Yellow: Personal Consumption Expenditures. Annual USD bio, from Bloomberg:

Today, that would imply a $1tr fall in spending. The frightening thing about the above chart however is what doesn’t move. The GFC barely impacted consumption. The same cannot be said for the current situation. Looking down the list of sectors there are some strong candidates for big hits.

USA GDP by expenditure, Q4 19 Annualised, in USD bio:

Car sales, Recreational goods, Transportation, Food services and Accommodation, Recreation - all of these categories will take a durable hit. Together, they’re worth $3tr. Lets haircut them by around 30% for the purposes of illustration. That leaves $2tr of spending that someone needs to do to offset the shock. Already, the $2tr CARES act looks appropriate instead of large. For a few reasons, it won’t be large enough.

The first problem, that we won’t get into in this short piece, is that $2tr number adds up a bunch of totally different things. 700bio of it funds loan guarantees - which isn’t spending at all. 500bio of it is going to states and government agencies, a good chunk of which will be spent on actual things. The remaining portion is cash transfers to individuals and companies - which the remainder of this piece will discuss. When I find a good explainer of the likely impact on government spending as opposed to headline grabbing exaggeration I’ll link back to it here. Lets give the benefit of the doubt though and take the $2tr figure as read. Congress has already expanded one element of the program by 300bio so it might even be fair to!

The problem for governments is that their spending mostly doesn’t buy output directly. Government consumption is about $3.5tr, with another $4tr in transfers. Mostly, they transfer income to other people - firms and households - who then buy things. Not all income is spent. First is the quarter or so of income that goes toward paying tax - so for every $1 the government hands out in income, they’ll replace 75cts or so of expenditure, so we need 2.7tr of income to replace our 2tr. Then, savings, which is where the real fun begins. Because everyone’s income is someone else’s expenditure, and only income can be saved, one individual’s desire to save reduces their expenditure, depriving someone else of the income they would like to save. This paradox of thrift explains how whilst economy wide savings rates do not fall in recessions, an increase in the desire to save drives down expenditure. For a government program of giving firms and individual more income, this means the bang for buck is reduced further as individuals and firms use incoming dollars to pay down debt.

The extent to which this is true is hard to determine - because frustratingly, economists prefer to look at historical events and try and tease out correlations between hard to understand macroeconomic aggregates than just survey households and see what they do with money. I managed to google up one such household survey done during a recession, that of 2001, where it was found that households spent about 2/3 of tax rebate money over 2 quarters. 2001 was a mild recession, but lets take it as representative and haircut our fiscal stimulus money again. Now, to get 2tr of expenditure, we need to provide 4.1tr of income. Put another way, if the US CARES bill is 2tr, we’d better hope the Covid19 shock to spending is less than 1tr - which is about the size of household expenditures on accommodation and food services. We’d also better hope that desired savings rates don’t rise as much as the mild producer side recession of 2001. I don’t like those odds. For this and other reasons my central scenario remains a wave of bankruptcies starting this summer as expenditure across the most impacted economies fails to recover.

PS:
If this sounds at odds with the Keynesianism we learnt in school, where we add up propensity to save and tax and import and calculate the fiscal multiplier - it’s because it is. To simplify massively, that theory assumes money and output are the same thing, which works for the most part when we’re dealing with economies in steady states. In a recession, we have to think differently. The purpose of replacing expenditure by firms and households is to prevent capital from being destroyed as the economy atrophies. Firms that don’t see demand don’t produce output, and shut down their productive assets (machines, premises etc). Extra spending is then required to get production going again, which means extra strain on balance sheets that are being hit by the recessionary impulse to tighten up. Preventing this is the crucial step to stopping a recession becoming a depression.

NB: This post is not investment advice and is not a trade recommendation. The views expressed here are my own and do not reflect those of my employer.