There’s been a lot of talk about how consumption will fall in the future — some of it added by myself [0]. I’m trying to fit some regressions now, to make some guesses about how consumption will move in the future, based on guesses about GDP and net wealth. I haven’t got very far, but at the very least, I can share some interesting pictures. Figure 1 depicts nominal shares of services, services and nondurables, and total (i.e., adding in durables) consumption, over the 1967-2008 period.

Figure 1: Services consumption to GDP ratio (blue), services plus nondurables consumption to GDP ratio (red) and total consumption (services plus nondurables plus durables) to GPD ratio (green). NBER defined recessions shaded gray. Source: BEA NIPA release of 30 October 2008, NBER, and author’s calculations.

What I find of interest is the long term upward trend in services consumption. Of course, this share is expressed in nominal terms, and we know that the relative price of many services has been rising.

The picture highlights the fact that nominal consumption has risen from about 60% of GDP in 1967 to about 70% in 2008. (It also reminds me that “stylized facts” change — when I was in graduate school, the ratio was about 65%; I’ll let readers guess when that was…)

In figure 2 I plot log prices of services, nondurables and durables. Notice that durables prices have been declining since the mid-1995s, and have started to stabilize.

Figure 2: log price deflator for services consumption (dark blue), nondurables consumption (pink) and durables consumption (dark green). Source: BEA NIPA release of 30 October 2008 and author’s calculations.

Perhaps more importantly, what is implied by this picture is that the relative price of durables has been declining quite rapidly over the past ten years (either vis a vis durables or services), and so looking at nominal expenditures can be misleading.

Unfortunately, it is hard to interpret the ratios of real consumption to real output expressed in Chained dollars (the ratio of two chained output series does not have a ready interpetation) [1]. Hence, I plot log values (note two vertical axes, with durables measured on the left hand side).

Figure 3: log real durables consumption to GDP ratio (dark green, left axis), log real services consumption to GDP ratio (dark blue) and log real nondurables consumption to GPD ratio (pink), all in Ch.2000$. NBER defined recessions shaded gray. Source: BEA NIPA release of 30 October 2008, NBER, and author’s calculations.

For the moment, I’ll observe that the trend behavior of durables consumption appears different from that of nondurables and services. We know from textbooks that’s true at the cyclical frequency (consumer durables are procyclical like investment), but here, we’ve seen a longer term trend. Hence, in any consideration of the future medium term path of consumption has to consider differential effects.

Hence, answering the question of what components of consumption will bear the greatest burden will require some view about the evolution of relative prices, as well as the intertemporal rate of substitution, and wealth effects.

Last 5 posts by Menzie Chinn

• Blogonomics: Some Random Thoughts - November 26th, 2009
• Debt and Interest Rates: Some Empirical Evidence and Implications - November 23rd, 2009
• Baselines, Counterfactuals and the Stimulus - November 21st, 2009
• China, the Renminbi, and Global Imbalances: A Quantitative View - November 20th, 2009
• GDP: Revisions and Forecasts - November 19th, 2009