In US Consumer – From King to Prince (8 October 2019), we argued that “the recent fall in US consumer confidence, slowdown in income and wage growth and jitters in US equity markets suggest that Personal Consumption Expenditure (PCE) growth remained weak in September and thus slowed materially in Q3” (September data are due on 31st October).
Weak September retail sales data, released on 16th October, support that view given the strong, positive historical correlation between the volume of retail sales and real PCE.
We estimate that the quarter-on-quarter SAAR of real growth in PCE halved to 2.3% from 4.6% which would imply that PCE growth in Q3 contributed only 1.6 percentage points to headline GDP growth, a little more than half of the 3.0pp it contributed in Q2 (see Figure 4).
The risk therefore is clearly biased towards GDP growth having slowed further in Q3 from 2.0% qoq in Q2. Note that the advance estimate of Q3 GDP is due out on 30th October at 13.30 (London time), only hours before the Federal Reserve concludes its policy meeting.
Forecasting PCE growth is complex as it is driven by a significant number of inter-related and often self-reinforcing factors. Yet our detailed analysis shows a strong, positive historical correlation between disposable income (including wages & salaries), US consumer confidence, households’ net-worth (including equity holdings), and PCE.
The slowdown in disposable income growth this year (due to weaker growth in “other incomes” and rising tax take), rising share of disposable income saved and recent fall in US consumer confidence do not bode well for PCE growth in Q4 and beyond.
However, the $7.2 trn rise in households’ net assets in H1 2019 and US equity markets’ resilience near record-highs should support consumer confidence and suggest that PCE and US GDP growth will remain positive, if modest, near-term.
This may not stop the Federal Reserve from cutting its policy rate 25bp once more before year-end, but the market’s current pricing of 30bp of rate cuts between now and-2019 looks a touch over-done, in our view.
Weak US retail sales in September supports our view that PCE growth under pressure
In “US Consumer – From King to Prince” (8 October 2019), we argued that:
1. “US Personal Consumption Expenditure (PCE) – or consumer demand – is critical to US economic growth. PCE growth has accounted for over 80% of US real GDP growth since end-2013 (see Figure 7), thanks to its relative size (nearly 70% of GDP) and the fact that PCE is the only demand sub-component which has positively contributed to GDP growth every quarter since 2010”.
2. “The recovery in real PCE growth in Q2 to 4.6% qoq (seasonally-adjusted annualised) was impressive but benefitted from weak growth in both Q4 2018 (1.4%) and Q1 2019 (1.1%). Indeed, month-on-month PCE growth gradually slowed from 0.8% in March to just 0.3% in July and 0.1% in August, with expenditure on services contracting in August – as also noted by the Federal Reserve in the minutes of its 18th September policy meeting.
3. “The recent fall in US consumer confidence, slowdown in income and wage growth and jitters in US equity markets suggest that PCE growth remained weak in September and thus slowed materially in Q3”. The US Bureau of Economic Analysis is due to publish personal expenditure and income data for September on 31st October.
Weak September retail sales data, released on 16th October, support that view. The volume of US retail sales has historically correlated closely with PCE and is thus a useful if imperfect lead indicator. Monthly data for the past 20 years show a high degree of correlation (R-squared of 0.72) between the growth rate of retail sales volumes and the growth rate in personal expenditure on goods (PCE on goods accounts for about 30% of total PCE), with no notable outliers (see Figure 1). Intuitively this makes sense.
The Dollar-value of total retail sales fell 0.25% mom in September, the first monthly fall since February although the August figure was revised to +0.6% mom from +0.4% mom (see Figure 2). Adjusting for inflation (i.e. in real terms), the picture is likely even bleaker. Assuming that headline PCE inflation rose 0.1% mom in September, real retail sales contracted 0.35% mom in September.
Retail sales were still up 4.0% yoy (in nominal terms) and 2.5% yoy (in real terms) in Q3, according to our estimates, but there has been a loss of momentum with the quarter-on-quarter seasonally-adjusted annualised rate (SAAR) of real retail sales growth slowing to 4.2% in Q3 from 5.1% in Q2 (see Figure 3).
Household consumption growth’s contribution to GDP growth likely halved in Q3
Based on the historical correlation between retail sales and PCE of goods, we estimate that the quarter-on-quarter SAAR of real growth in the PCE of goods slowed to 4.8% in Q3 from 8.4% in Q2 and that growth in total PCE halved to 2.3% from 4.6% (see Figure 4). This would tie with the Federal Reserve’s observation on 18th September that recent data suggested “some slowing in consumer spending growth in the third quarter from its strong second-quarter pace”.
If our forecast proves correct, it would imply that PCE growth in Q3 contributed only 1.6 percentage points to headline GDP growth, a little more than half of the 3.0pp it contributed in Q2 (see Figure 4). At the same time growth in fixed investment and government spending has in the past 18 months been comparable to overall GDP growth while inventories and net exports have been a drag (see Figure 5).
We conclude that the risk is that GDP growth may have slowed further in Q3 from 2.0% qoq in Q2 (see Figure 6), with the Atlanta Fed currently estimating that Q3 GDP growth was 1.8% qoq in Q3. A further slowdown in GDP growth would put the Federal Reserve under pressure to cut its policy rate 25bp again before year-end. Note that the BEA will release its advance estimate of Q3 GDP on 30th October at 13.30 (London time), only hours before the Federal Reserve concludes its two-day policy meeting and makes its announcement at 19.00.
Drivers of household consumption: Income, confidence and wealth
Modelling PCE growth is complex as it is driven by a significant number of inter-related and often self-reinforcing factors. According to the Federal Reserve, these include the unemployment rate, real disposable income, households’ net worth (the “wealth effect”) and borrowing rates. Arguably, the unemployment rate and disposable income are inter-connected to the extent that a low level of unemployment (and high level of employment) is likely to be associated with higher aggregate wages and thus disposable income. Moreover, we would argue that consumer confidence is an important factor behind PCE growth, as we summarise in Figure 7.
The Federal Reserve in the minutes of its 18th September policy meeting argued that “Key factors that influence consumer spending – including a low unemployment rate, further gains in real disposable income, high levels of households’ net worth, and generally low borrowing rates – were supportive of solid real PCE growth in the near-term.” Weak retail sales data for September arguably challenge this view and our analysis suggests that PCE growth may remain modest in Q4.
1. Disposable income growth has slowed
The most obvious driver of PCE growth is arguably the growth in disposable income and the share of disposable income that is consumed, saved and spent on other outlays (personal interest payments and current transfer payments). Figure 8 indeed shows a strong, positive historical correlation between the year-on-year growth rate of real disposable income growth and real PCE growth. Both have slowed in tandem in the past year.
Since mid-2005, PCE growth has tended to lag disposable income growth, which we attribute in part to the increase in the share of disposable income which has been saved (rather than spent) from about 2.5% to 8% (see Figure 9) and the commensurate fall PCE as a share of disposable income from about 94% to 88.5%. The share of disposable income spent on other outlays has been broadly constant around 3.5% (see Figure 10). To put this in context, had PCE’s share of disposable income remained constant at 94%, PCE (in nominal terms) in the past 12 months would have been $15.25trn or $887bn higher than it actually was – broadly equivalent to the Netherland’s annual GDP.
Disposable income itself is the product of wages & salaries and other income – which each account for about half of personal income – minus taxes (see Figure 10). Other income is composed of supplements to wages & salaries, proprietors’ income, income from rent, savings and dividends, and benefits.
Real year-on-year growth in wages & salaries has been stable around 4% since February (see Figure 11), as already noted in Retail sales to the rescue (29 April 2019). However, based on Bureau of Labour CES data for aggregate weekly payrolls in September and assuming PCE inflation of 0.1% mom, we estimate that wages and salaries (in constant prices) were broadly unchanged from August. The relationship between wage growth and PCE arguably works both ways, to the extent that slower PCE growth could make US corporates more reluctant to award workers higher wages & salaries.
Moreover, growth in other sources of income has gradually slowed in the past 12 months while at the same time growth in taxes has risen slightly (see Figure 11). Indeed, taxes as a percentage of personal income have risen from about 11.4% at end-2018 to about 11.8%.
2. Consumer confidence key to consumption growth
Our analysis also suggests that flat-lining consumer confidence since end-2018, as measured by the Conference Board, has weighed on PCE growth (see orange box in Figure 12). Indeed, Figure 12 shows a strong, positive historical correlation between consumer confidence and the real year-on-year growth rate in PCE, albeit with a temporary dislocation between mid-2016 and March 2017 highlighted by the green box. Put differently, US households need to be confident about the economic outlook to spend (rather than save) any increase in personal income – a point we made in Retail sales to the rescue (29 April 2019).
At the same time, the savings rate itself can impact consumer confidence to the extent that a high savings rate – as is currently the case – can, along with other household assets, positively impact consumer confidence as we discuss in the following section. This may explain in part why US consumer confidence remains high by historical standards (125.1 in September).
3. Wealth effect, including resilient equities, supporting PCE and consumer confidence
US households’ net worth (the “wealth effect”) is also an important direct and indirect factor behind PCE growth, as noted by the Federal Reserve in recent minutes, including those of its 18th September policy meeting.
Household’ net worth are its assets – equity holdings, real estate, consumer durables (e.g. cars), deposits and other financial assets including savings – minus its liabilities (including mortgages and other loans). Net assets were up about 5% yoy at end-June 2019 to $113.5 trn thanks to rising stock and real estate prices (see Figure 13), with the $5.3trn increase in Q1 2019 the largest on record. Small percentage changes in households’ net assets can therefore have a material impact on the level of net assets and households’ perception of their net-worth and therefore on PCE.
Figure 15 indeed shows a strong, positive historical correlation between the percentage change in households’ net assets and real PCE growth, with PCE lagging households’ assets by about two quarters. When households’ net worth increases, they have – all other things being equal – greater means to spend, even if some of those assets (e.g. real estate) are not particularly liquid.
Direct holding of stocks and indirect holding of stocks (through 401(k)-type retirement plans) account for respectively 11% and 16% of households’ net assets and are equivalent to respectively 74% and 110% of households’ disposable personal income (see Figure14). Small changes in equity prices can thus have a material impact on households’ net assets. Moreover, households can more easily track the value of their stock holdings and liquidate them, particularly those directly held, compared to harder-to-price, less liquid assets (such as real estate). Therefore, changes in stock prices are likely to have a more direct and immediate impact than other assets on PCE growth, consumer confidence and even wages, in our view.
When US equities are rising, households – of which about 52% directly and indirectly own US equities – have greater wealth to spend. Figures 16 and 17 indeed show a strong, positive historical correlation between the level and year-on-year percentage change in the S&P 500 and the nominal $-value of PCE.
At the same time, if PCE growth is robust, US companies (selling into the domestic market) are more likely to perform well and see their share price rise, which bumps up the value of households’ stock holdings and net assets. If US companies’ stock price is rising and more confident US consumers are spending some of their increased wealth, companies are more likely to be in a position to increase wages, which in turn boosts consumer confidence and spending. There is indeed a strong, positive historical relationship between the S&P 500 and consumer confidence (see Figures 18 & 19). It is no coincidence, in our view, that US President Trump regularly talks up the near-record high in US equities and is keen for monetary policy to continue stimulating higher equity prices.
If these historical relationships and virtuous cycle between stock prices, PCE, consumer confidence and wages hold going forward, the rebound in households’ net worth in H1 2019 and the S&P 500’s resilience should provide some near-term support to household consumption and ultimately keep real PCE and US GDP growth in positive territory, in our view.
Put differently, talk of a US recession – two consecutive quarters of negative GDP growth – seems very premature and, with this in mind, we think the market’s current pricing of 30bp Fed rate cuts between now and end-year is a touch over-done. The risk to our core scenario that the Federal Reserve does not cut rates again before year-end is admittedly on the dovish side given the recent spate of weak US macro data, including retail sales, ISM PMIs, Philadelphia Fed manufacturing index, industrial output, weekly earnings growth and non-farm payrolls. However, we think that the Federal Reserve is more likely to keep rates on hold for the remainder of the year than to cut rates twice more.
However, should US equities suffer a significant contraction, the negative impact – both direct and indirect – on household consumption growth and therefore US GDP growth would materially increase the odds of the Federal Reserve delivering one and potentially two more 25bp rate cuts this year.
 “Real PCE rose briskly in July, while the components of the nominal retail sales data used by the Bureau of Economic Analysis (BEA) to estimate PCE were flat in August and the rate of sales of light motor vehicles only edged up”.