Productivity growth has been puzzlingly weak in the wake of the Great Financial Crisis (Hughes and Saleheen 2012). This has heightened concerns about secular stagnation – the view that a structural deficiency of demand has been haunting the world and crippling its productive capacity as persistent unemployment erodes workers’ skills and investment slumps (Summers 2014, Teulings and Baldwin 2014). Those who hold this view typically also argue that the outsize financial booms – unsustainable increases in credit and property prices – that preceded the crisis were the price to pay for keeping resources fully employed. Once the booms turned to bust, the hidden demand deficiency re-emerged with a vengeance.
This narrative has some merit and has gained currency. But what if, in addition to the persistent – but not structural – hole in aggregate demand a financial bust inevitably generates, a key part of the true story has less to do with the level of aggregate demand than with its composition and impact on the structure of supply? What if what some see as a rather disappointing pre-crisis US growth performance despite a strong financial boom was actually disappointing, in part, precisely because of that boom? What if the protracted post-crisis weakness reflects in no small measure the difficulties in correcting the resource misallocations that accumulated during the previous financial boom and emerged once a financial crisis subsequently broke out?
This is indeed what we conclude by examining the experience of 21 advanced economies over the last 40 years (Borio et al. 2015). The hitherto unsuspected villain in this story is the misallocation of resources – in our case, labour – during the credit boom and its long post-crisis shadow. More generally, the findings support the view that the disappointing developments we have been witnessing may be the result of a major financial boom and bust that has left long-lasting scars on the economic tissue (e.g. BIS 2014, Borio 2014, Borio and Disyatat 2014, Rogoff 2015) rather than the reflection of a structural, deep-seated weakness in aggregate demand.
What do we find?
Figure 1 Financial booms sap productivity by misallocating resources
Notes: Estimates calculated over the period 1969–2013 for 21 advanced economies. 1 Annual impact on productivity growth of labour shifts into less productive sectors during the credit boom, as measured over the period shown. 2 Annual impact in the absence of reallocations during the boom.
Source: Based on Borio et al. (2015).
Figure 1 summarises our key findings. To help fix ideas, it shows the impact on productivity of a synthetic credit boom-cum-financial crisis episode – specifically, the impact of an assumed five-year credit boom followed by a financial crisis, and considering a five-year post-crisis window. Three points stand out.
First, credit booms tend to undermine productivity growth as they occur. For a typical credit boom, a loss of just over a quarter of a percentage point per year is a kind of lower bound.
Second, a large part of this, slightly less than two thirds, reflects the shift of labour to lower productivity growth sectors – this is the only statistically significant component. Shifts into a temporarily bloated construction sector play a key role. The remainder is the impact on productivity that is common across sectors, such as the shared component of aggregate capital accumulation and of total factor productivity.
Third, the subsequent impact of labour reallocations that occur during a boom is much larger if a crisis follows. The average loss per year in the five years after a crisis is more than twice that during the boom, around half a percentage point per year. The reallocations cast a long shadow.
Regardless of the specific figure, the overall impact is sizeable. Taking the ten-year episode as a whole, the cumulative impact amounts to a loss of some 4 percentage points. Put differently, and focusing on the countries that experienced a banking crisis in 2007–08, these findings suggest that the loss in productivity growth over 2008–2013 as a result of the preceding credit boom was roughly as large as the actual productivity growth over the same period (around 0.6%). Think, for instance, of the painful rebalancing in the structure of production that Spain had to go through following the pre-crisis boom, as confirmed by other empirical evidence (e.g. García-Santana el al. 2015).
How do we find it?
To uncover these findings, we proceed in three steps.
First, we borrow from Olley and Pakes (1996) a decomposition of productivity growth – purely an identity – into a common and an allocation component, but apply it across sectors rather than within sectors, unlike what the authors originally did. We consider nine sectors. To give some orders of magnitude, in our sample over the period 1979–2009 the allocation component on average accounts for around one third of labour productivity growth.
Second, we regress each component in turn, calculated over non-overlapping three- or five-year windows, on measures of a credit boom and a series of variables intended to control for the influence of other factors. The results indicate that the only economically and statistically significant relationship is between credit and the allocation component (compare the left-hand panel with the right-hand panel of Figure 2). In other words, not only do credit booms undermine productivity growth, as already found by Cecchetti and Kharroubi (2015), but they do so mainly by inducing shifts of resources into lower productivity growth sectors.
Figure 2 Financial booms and productivity growth components
(Computed over five-year windows and taken as deviations from country and period means)
Notes: The panels plot the growth rate in private credit to GDP against the allocation and common components of labour productivity growth, respectively; both variables are taken as deviations from country and period means. The sample includes 21 economies (Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, Korea, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom and the United States) and six periods of five years (1979–84; 1984–89; 1989–94; 1994–99; 1999–2004; 2004–09).
Source: Authors’ calculations.
Finally, we examine how well the behaviour of the two productivity components during credit booms, and economic expansions more generally, predicts the behaviour of productivity during the subsequent recessions and their aftermath. Here, too, we control for the influence of other factors and examine, in particular, how the evolution of productivity depends on whether a financial crisis occurs or not. A key result is that the predictive content of the reallocation component is much greater if a crisis erupts.
The size and persistence of the overall productivity slowdown when a financial crisis erupts and labour misallocations were relatively high during the boom are shown in Figure 3. The results suggest that this configuration can lead to long-lasting productivity stagnation: productivity may return to its previous level only some eight years after the GDP peak.
Figure 3 Productivity stagnates after a financial crisis due to previous labour misallocations
Notes: This simulation is based on local projection regressions of the percentage deviation of labour productivity from the recession year. The independent variables include the allocation and the common components of productivity growth over the three-year period prior to the start of the recession. The solid blue line shows the projection of labour productivity conditional on the occurrence of a financial crisis and a positive allocation component contribution of 0.85 percentage points over three years prior to the recession (first quarter of the distribution of the allocation component contribution). The blue area around the blue line represents the 5% confidence interval around the projected productivity path.
Source: Authors’ calculations.
What do the findings indicate?
We do not provide a formal model to interpret these findings; even so, possible mechanisms come to mind. These involve, during credit booms, the different incidence of credit expansion across sectors, not least owing to collateral characteristics. And they involve, during busts, the interaction between financial crises, the scarcity of credit, slow balance sheet repair and the need to reverse the previous resource misallocations linked to temporarily bloated sectors. Under these conditions, misallocations may beget misallocations. These issues deserve deeper scrutiny.
Our analysis raises broader questions, well beyond those that concern the interpretation of the post-crisis slowdown in productivity growth and the explanatory power of the secular stagnation hypothesis.
First, it suggests that when considering the macroeconomic implications of financial booms and busts, it is important to go beyond the well known and very real aggregate demand effects and to examine also what happens on the supply side of the economy. Thus, our findings help explain the usefulness of financial cycle proxies, notably credit and property prices, in the real-time measurement of potential output during the boom (Borio et al. 2013). They also provide a complementary explanation for hysteresis effects – one linked to the allocation of credit and the costs of reallocating real resources rather than simply to protracted aggregate demand weakness. All this highlights the importance of supply side policies.
Second, the findings enrich our understanding of the medium- to long-run effects of monetary policy and of its effectiveness in addressing financial busts (Borio 2015). If loose monetary policy contributes to credit booms and these booms have long-lasting, if not permanent, effects on output and productivity, then it is not reasonable to think of money as neutral over long-term policy horizons. After all, financial booms and busts linked to crises have had a length of between 16 and 20 years (e.g. Drehmann et al. 2012), and our results confirm that misallocations take time to develop and have very long-lasting effects. Nor is it surprising if monetary policy may not be particularly effective in addressing financial busts. This is not just because its force is dampened by debt overhangs and a broken banking system. It may also be because loose monetary policy is a blunt tool to correct the resource misallocations that developed during the previous expansion, as it was a factor contributing to them in the first place. All this adds strength to the view that there is a case for monetary policy to lean against financial booms.
Finally, the findings underline the need to use a wide range of models in monetary policy analysis. In policy formulation, all too often the conclusions are based on the standard ‘one-good’ benchmark model – that is, on models that behave as if there were only one good. The models used in policymaking should be able to accommodate the implications of costly sectoral shifts, well beyond those linked to the time-honoured distinction between tradables and non-tradables. Otherwise there is a risk of throwing out the baby with the bathwater.
Authors’ note: The views expressed are those of the authors and do not necessarily represent those of the Bank for International Settlements.
Bank for International Settlements (2014), 84th Annual Report, June.
Basel Committee on Banking Supervision (2010), An assessment of the long-term economic impact of stronger capital and liquidity requirements, July.
Borio, C (2014), “The financial cycle and macroeconomics: what have we learnt?”, Journal of Banking & Finance, 45, August, pp 182–98. Also available as BIS Working Papers, no 395, December 2012.
Borio, C (2015), “Revisiting three intellectual pillars of monetary policy received wisdom”, Cato Journal, forthcoming. Also available under BIS Speeches at http://www.bis.org/speeches/sp151112.htm.
Borio, C and P Disyatat (2014), “Low interest rates and secular stagnation: is debt a missing link?“, VoxEU.org, 25 June.
Borio, C, P Disyatat and M Juselius (2013), “Rethinking potential output: embedding information about the financial cycle”, BIS Working Papers, no 404, February.
Borio, C, E Kharroubi, C Upper and F Zampolli (2015), “Labour reallocation and productivity dynamics: financial causes, real consequences”, BIS Working Papers, no 534, December.
Cecchetti, S and E Kharroubi (2015), “Why does financial sector growth crowd out real economic growth?”, BIS Working Papers, no 490, February.
Dias, D, C Robalo Marques and C Richmond (2014), “Misallocation and productivity in the lead up to the Eurozone crisis”, Banco do Portugal Working Papers, no 11.
Drehmann, M, C Borio and K Tsatsaronis (2012), “Characterising the financial cycle: don’t lose sight of the medium term!”, BIS Working Papers, no 380, November.
García-Santana, M, E Moral-Benito and J Pijoan-Mas (2015), “Growing like Spain: 1995–2007”, working paper, Universitat Pompeu Fabra, Department of Economics.
Gopinath, G, S Kalemli-Özkcan, L Karabarbounis and C Villegas-Sánchez (2015), “Capital allocation and productivity in South Europe”, CEPR Discussion Paper No 10826.
Hughes, A and J Saleheen (2012), “UK labour productivity before and after the crisis – an international and historical perspective”, Bank of England Quarterly Bulletin, 52(2), pp 138–46.
Hsieh, C-T and P Klenow (2009), “Misallocation and manufacturing TFP in China and India”, Quarterly Journal of Economics, 124(4), pp 1403–48.
Jordà, O, M Schularick and A Taylor (2013), “When credit bites back”, Journal of Money, Credit and Banking, supplement to 45(2), pp 3–28.
Olley, S and A Pakes (1996), “The dynamics of productivity in the telecommunications equipment industry”, Econometrica, 64(6), pp 1263–98.
Rogoff, K (2015): “Debt supercycle, not secular stagnation”, VoxEU.org, 22 April.
Summers, L (2014), “Reflections on the ‘New Secular Stagnation Hypothesis’”, in C Teulings and R Baldwin (eds), Secular stagnation: facts, causes and cures, VoxEU.org eBook, CEPR Press.
Teulings, C and R Baldwin (2014): Secular stagnation: facts, causes and cures, VoxEU.org eBook, CEPR Press.
 Building on the seminal work by Hsieh and Klenow (2009), who consider the within-sector dispersion of marginal productivity a sign of misallocations, two recent pieces of research shed light on the Spanish and Portuguese experience, finding large misallocations during the boom (Dias et al 2014, Gopinath et al. 2015).
 Specifically, we use local projection methods (eg Jordà et al (2013)).