Entries from August 30, 2020 - September 5, 2020
A "monetarist" inflation forecast
G7 annual broad money growth was little changed at 16.9% in June. Recent growth is the fastest since 1973 – the post-WW2 peak was 18.3% in November 1972. The average in the post-GFC decade (2010-19) was 3.7%.
The monetarist view is that the monetary surge in 2020 will be reflected in a bulge in nominal GDP growth and inflation in 2021-22 as the velocity of circulation normalises following an involuntary plunge caused by government restrictions on economic activity.
The magnitude of the inflation rise and the extent of a subsequent reversal will depend on how far and fast money growth falls back from the current extreme.
On reasonable assumptions about velocity and money growth, G7 inflation could average 4-5% in 2021-22, as explained below.
G7 broad money velocity (i.e. nominal GDP divided by the money stock) trended lower over 1964-2019 but the rate of decline increased between the two halves of this period – see first chart. A simple approach is to model velocity as a split time trend with a break at the mid-point of the period (i.e. end-1991).
Actual velocity was close to this trend in Q4 2019 but the covid crash in nominal GDP resulted in a 19% undershoot in Q2 2020. The forecast below assumes that 1) velocity returns to trend and 2) the trend rate of decline remains constant (1.75% pa).
Multiplying the money stock by trend velocity generates an implied level of nominal GDP towards which the actual level, on the monetarist view, is likely to gravitate. The historical path of this implied level is shown in the second chart along with a projection based on broad money rising at a constant 4% annualised rate from its July level, i.e. close to the post-GFC average.
Assume now that actual nominal GDP converges with the implied level in Q4 2022. This would imply a 17% rise in nominal GDP from its pre-covid level in Q4 2019 in the three years to Q4 2022.
Real GDP is expected to return to its pre-covid level by early 2021 and may continue to grow strongly into 2022 but the cumulative rise over Q4 2019-Q4 2022 is unlikely to exceed 6%. A 17% increase in nominal GDP, therefore, could imply a 10-11% rise in prices, as measured by the GDP deflator, in the three years to Q4 2022.
Monetary changes usually take at least year to be reflected in prices so inflation is likely to remain low in H2 2020. The GDP deflator could rise by about 1% between Q4 2019 and Q4 2020. The suggested 10-11% increase in the three years to Q4 2022 would then imply a rise of 9-10% between Q4 2020 and Q4 2022, i.e. an inflation rate of 4-5% pa.
An unsatisfactory feature of the above analysis is the lack of an economic explanation for the falling trend in velocity. The rate of decline, in theory, could speed up, offsetting the inflationary impact of the 2020 monetary surge.
Such an explanation was provided in previous posts on the “quantity theory of wealth”. The demand to hold money depends on the level of wealth as well as nominal income. A rise in the wealth to income ratio explains most of the secular decline in velocity.
The empirical analysis also suggests a role for real government bond yields – the fall in real yields since the 1980s may explain the faster rate of decline of velocity in the second half of the 1964-2019 period noted earlier. (This is consistent with Keynes's "speculative" motive for holding money: “high” yields cause some investors to expect a fall and consequent capital gain, leading them to switch from money to bonds; conversely, “low” yields generate expectations of capital loss so increase money demand.)
The third chart compares actual G7 broad money velocity with the fitted values of a simple model based on these two factors, i.e. the G7 wealth to income ratio and 10-year real government bond yields. Like the split time trend model discussed earlier, this model suggests an unprecedented divergence between actual velocity and its “equilibrium” level.
Since real bond yields are unlikely to repeat their decline since the 1980s, the model implies a slowdown in the trend fall in velocity unless the wealth to income ratio rises even faster than in recent decades.
Don't call it a recession
The National Bureau of Economic Research (NBER) business cycle dating committee, the arbiter of US recession / expansion chronology, stated in June that a peak in monthly economic activity had occurred in February 2020, implying the start of a recession. This determination was premature and unwise. Monthly data indicate that the “recession” ended in April, implying that it was the shortest in history by a significant margin and did not fulfil the NBER’s traditional duration requirement.
The issue is not merely semantic. The recession determination is likely to have influenced investor behaviour, causing some market participants to reduce exposure to economy-sensitive assets and thereby suffer a significant opportunity cost from the V rebound in markets.
According to the NBER’s traditional definition, “a recession involves a significant decline in economic activity that is spread across the economy and lasts more than a few months”. The duration requirement captures the self-feeding nature of recessions, in which an initial negative impulse is magnified and extended by multiplier / accelerator effects.
There was no such internal development in the recent output contraction, which was caused solely by a government decision to place controls on economic activity for health reasons and reversed when these were relaxed. The brevity of the restrictions coupled with government / central bank support for business and consumer cash flows weakened second-round effects.
The NBER’s monthly assessment places particular emphasis on two indicators: real personal income excluding transfer payments and payroll employment. Both bottomed in April and recovered significantly through July.
The Conference Board’s coincident economic index comprises the above two indicators along with industrial output and real business sales. The index fell by 13.6% from a February peak to an April low but rose by 6.7% between April and July.
The NBER, it seems, will be forced to concede that the “recession” ended in April, implying that it has already been exceeded in duration by the new “expansion” phase.
The NBER’s recession chronology extends back to 1854. A two-month recession would be the shortest in history by a significant margin – the previous minimum was six months, i.e. January-July 1980. The latter recession, similarly, reflected government action – the imposition of credit controls – and ended when these were removed.
The 1980 recession was followed by another deeper and longer contraction starting 12 months later in July 1981. It has been suggested that labour market and credit damage from the covid shock will result in a similar “double dip” in late 2020 / early 2021. The 1981-82 recession, however, was caused by Fed Chair Volcker’s assault on double-digit inflation – the Fed funds rate rocketed from 11% to 19% between August 1980 and May 1981. Monetary policy will remain super-loose for the foreseeable future, although fiscal policy could tighten significantly next year.
The current NBER committee has attempted to justify its disregard of the duration requirement, arguing that “the unprecedented magnitude of the decline in employment and production, and its broad reach across the entire economy, warrants the designation of this episode as a recession, even if it turns out to be briefer than earlier contractions”. It is doubtful whether previous distinguished committee members who instituted the “depth, diffusion and duration” necessary conditions would have agreed with this shifting of the goalposts.
The recession determination may be feeding into investor expectations of a “normal” economic recovery in which capacity remains underutilised and inflationary pressures weak for a prolonged period. The forecast here is that an ongoing V rebound in activity will develop into a full-scale boom in 2021, resulting in supply-side bottlenecks, commodity price strength and a surprisingly rapid reversal of employment losses.