Winner of the New Statesman SPERI Prize in Political Economy 2016

Tuesday 2 December 2014

Secular stagnation and computers

For macroeconomists

Secular stagnation means different things to different people, but a common motivation is the steady decline in real interest rates since the 1980s. There have been many explanations for this (some of which I discuss here), and a common feature is that while some mechanisms are quite plausible (in particular a reduction in population growth will reduce real rates in most models) individually they do not seem quite enough. It therefore appears likely that we could be looking at something with multiple causes. So here is another possible mechanism that can be added to the list, which is the fall in the price of new investment goods.

Although this link between investment goods prices and secular stagnation has been suggested before, I want to focus on a new paper by Gregory Thwaites of the LSE and Bank of England [1], which explores this effect in a complete model. His paper has some similarities to the paper by Eggertsson and Mehrotra that I talked about in this post (for example it uses a three period OLG model), but it has much more of a focus on this investment goods price effect.

One of the very striking features of recent decades has been the relatively slow growth in the price of investment goods compared to the more familiar price of consumption goods. As Karabarbounis and Neiman (2014) [2] note, this is often attributed to advances in information technology and the computer age. Thwaites extends Karabarbounis and Neiman’s data across countries and time, and shows that this decline in the price of investment goods occurs across countries, and did appear to begin around 1980.

A key issue is how much firms react to the fact that capital is becoming cheaper by substituting capital for labour. In Karabarbounis and Neiman they react with an elasticity of substitution greater than one, and they use their analysis to explain a decline in the labour share. However, as Thwaites notes, there are many studies which suggest a less than unit elasticity of substitution.

To see the implications of this, consider a very simple OLG model with log consumption and where agents only work in the first period. This implies that the proportion of income saved is constant. So if the fall in the price of investment goods leads, ceteris paribus, to a fall in the value of capital required by firms, then to equate the demand and supply for savings real interest rates will fall. (You need an OLG framework here. In the benchmark representative agent model, the real interest rate equals the rate of time preference plus the growth rate.) This is a steady state result, and the paper explores the dynamics.

That is the key idea. For me, a really interesting aspect of the paper is that it integrates housing into this analysis. If real interest rates fall, and for whatever reason the supply of housing is fixed, house prices will rise (see these two posts). This leads to an increase in gross household debt, because agents borrow from the old to buy houses. Thwaites’s model shows that this dampens the fall in real interest rates, because this is an alternative destination besides capital for retirement savings.

I recognised the mechanism, because I had been playing around with the same effect when looking at steady state changes in government debt. A permanent reduction in the ratio of government debt to GDP would in an OLG model free up savings for capital, reducing real interest rates (as we explored in this paper for example). But lower real rates also raise the demand for housing, which can be an alternative way of saving for retirement. More generally, whatever the causes of this apparent trend decline in real interest rates, the implications for the housing market - and what we think of as ‘normal’ in that market - are likely to be profound. 

[1] Thwaites, G (2014) Why are real interest rates so low? Secular stagnation and the relative price of investment goods, Centre for Macroeconomics Discussion Paper No. CFM-DP2014-28

[2] Karabarbounis, L. and B. Neiman (2014). The global decline of the labor share. The Quarterly Journal of Economics 129 (1), 61–103.


  1. Neat! Just the sort of model I was looking for. Thanks for this post.

    When they say "houses", they really mean "the land on which houses are built", right?

    Pity there isn't a simple way to explore the effects of increasing longevity in retirement in an OLG model. Or is there? You want to make the third (last) period longer, but that doesn't work. Maybe have fewer people die before getting to the third period?

    1. And it bugs me hearing people talk about "house price bubbles", without seeing the rise in house/land prices as just one symptom of the fall in real interest rates. The yield (rent/price ratio) on houses has been falling along with yields on other assets.

    2. Nick,

      I think you could do this by allowing for each individual in a given generation to face some probability of death at the end of the second period (in addition to certain death at the end of the third), and then changing this probability over time. Is that what you meant? Saving could then take the form of buying insurance against the possibility of surviving into the third period (not unlike actual pension schemes). I think the effect would be similar to varying the rate of time preference for the third period.

    3. “When they say "houses", they really mean "the land on which houses are built", right?”

      According to The Economist house price index, the price of houses in the UK in real terms has doubled over the last 20 or 30 years. That’s accounted for TO SOME EXTENT by a rise in land with permission on which to build houses. But it’s not the whole story. This Policy Exchange study has something on land price rises:

      And just to muddy the whole issue, according to the Economist, the real price of houses in Germany, Switzerland and Japan has remained constant over the last 20 or 30 years.

      I’m baffled.

    4. Nick E: Yes! That's exactly what I meant. Plus, if we let people choose to work part-time when they are old (and choose the number of hours) that is roughly like making the retirement age endogenous. That way we could see how the increased lifespan affects the desired stock of saving, with the retirement age (sort of) endogenous.

      Ralph: that is puzzling. Because bricks and mortar should have a very elastic supply curve. Maybe it's just that the technology for building (Brit-style) houses has not improved much over the years, relative to producing other goods. Very traditional when it comes to houses, they are. Plus they like old houses more than new, and old houses are in inelastic supply.

    5. According to Statistisches Bundesamt in Germany the index of prices of houses (Hauspreisindex) has risen by 10% from 2000 to 2013.

  2. "Secular stagnation means different things to different people, but a common motivation is the steady decline in real interest rates since the 1980s."

    Does this make any sense? According to the World Bank data, real interest rates in the UK in 1967 were 2.5%. In the 1970s, they were sometimes negative. From 1982 to 2000, they ranged between 4 and 6% (pretty high.) They fell under 3% in 2001. After 2009 they have been negative.

    So here is the explanation for the patterns. In 1967, rates were what you see in a normal functioning economy. In the 1970s, spikes in inflation from oil price shocks caused inflation to surpass even very high interest rates, causing real rates to become negative.

    From 1979 to 1982, extremely high interest rates "broke the back of inflation." But after 1982, interest rates remained high due to continued inflation fighting (monetarism) which caused real interest rates to get very high by post-war standards.

    From 1989 to 2000, real rates remained high because of more inflation fighting (2% inflation target.) This "Great Moderation" was short-lived. It brought back boom-to-bust economic cycles. First the dot-com bubble of the late 1990s burst causing the 2001 recession. After that there was the derivatives bubble that caused the 2008 global economic meltdown.

    In short, tight money was responsible for the very high real interest rates from 1982-2000. And it was responsible for successive economic collapses that caused real interest rates to fall as GDP growth plummeted: the 2000s was a miserable decade for growth — the worst since the Great Depression, by far. The 2010s will, of course, be a much greater mess — despite copious amounts of helicopter money dropped on the heads of rich bankers.

    (Japan has been stuck in a similar slump for 20 years. Of course, that won't happen in Western countries because "academic economists" have a superior knowledge of monetary theory — despite having caused all the mess in the first place.)

  3. "If real interest rates fall, and for whatever reason the supply of housing is fixed, house prices will rise (see these two posts). This leads to an increase in gross household debt, because agents borrow from the old to buy houses."

    I think economists really have to start asking themselves if they are getting anywhere with their neoclassical economic models. First the assumptions are out of touch with reality. Second they are overly simplistic ignoring complex historical realities. Third, if an economist plays with the switches and dials and, by chance, comes up with some particular finding that vaguely resembles reality, (ignoring everything that doesn't,) all it does is reinforce his or her belief system that the model works, when its really garbage in, garbage massaged and manipulated, and garbage out.

    Economists with physics envy should try googling the "three body problem." Even when the laws that govern a system are known, the results become unpredictable in a simple 3 body system. As Keynes said of the economy (which, if the results are any indication, is more true than ever): it's "a delicate machine, the working of which we do not understand."

    It's it really any surprise that very low interest rates are fueling housing bubbles in various countries (that didn't already suffer a housing collapse)? Will it be any surprise when this ends in yet another disaster?

  4. Another factor leading to the slow decline in interest rates will presumably be that savers / creditors who got their fingers burned in the 1970s inflation have long memories. I.e. during the 80s and 90s they wouldn't have been willing to lend other than at a rate that compensated them for a resurgence of inflation.

    1. Central banks set interest rates, not savers/creditors. Their heavy-handed inflation-fighting monetary ideology caused the high real interest rates of the 1980s and 1990s. It also brought back boom-to-bust economic cycles.

      Central banks are now forced to hold interest rates near zero to prevent a deflationary spiral. But this helicopter money has obviously done little to stimulate a recovery, 6 years and counting.

      In fact, the US Fed was forced to lower interest rates to 1% in the early 2000s. Back then the hull had scraped bottom. By the end of the decade, the self-proclaimed technocrats had run the ship aground.


Unfortunately because of spam with embedded links (which then flag up warnings about the whole site on some browsers), I have to personally moderate all comments. As a result, your comment may not appear for some time. In addition, I cannot publish comments with links to websites because it takes too much time to check whether these sites are legitimate.