When we were college students in the mid-1970s, some of our friends wanted to change the world and our understanding of it. They worked on things like galactic structures, superconductors, computer algorithms, subcellular mechanisms, and genetic coding. Their work aimed at providing cheap energy, improving information technology, curing cancer, and generally making our lives and our appreciation of the world around us better.
By the 1990s, attitudes had changed. Many top students, including newly-minted Ph.D.s, moved from natural science and engineering to finance. Their goal was to get high-paying jobs.
Would we be better off today if some of these financial wizards had focused instead on inventing more efficient solar cells or finding ways to forestall dementia? The older we get, the more we think so (especially when it comes to dementia). And, believe it or not, there is now notable, cross-country evidence buttressing this view.
For the past few years, one of us (Steve) has been studying the relationship between economic growth and finance. The results are striking. They come in two categories. First, the financial system can get too big – to the point where it drags productivity growth down. And second, the financial system can grow too fast, diminishing its contribution to economic growth and welfare.
Just to get this on the record: we really like finance. Without efficient financial services, there would be no prosperous economies today. Intermediaries and financial markets both mobilize and channel savings to those who can use capital most productively; they also allocate risk to those persons who are most able to bear it. Despite its recently tarnished reputation, financial innovation greatly improves people’s lives (see here). As a result, when finance is properly harnessed, it makes economies more productive, enhancing employment and growth, and makes the world a better place.
So, we badly need efficient finance. But how much do we really need? And, should we be concerned if financial sector growth sharply outpaces the growth in the rest of the economy?
The following two charts address these two questions. The first shows the relationship between the size of the financial sector – measured as the average share of total employment – and productivity growth. Each dot represents a five-year episode in a specific country. The data are from a set of advanced (OECD) economies over a 30-year period. (The plot is a bit complicated to draw, as it removes country-specific effects and differences in initial conditions.)
While there is admittedly a lot of noise in the pattern (including some significant outliers), the results suggest that finance can get too large. Put differently, as the share of employment in the financial sector increases, observed productivity growth first increases and then decreases. The measured peak corresponds to a sectoral employment share of about 4% – a level that a number of countries clearly exceeded in the past, and some still do. [In the United States, February 2015 employment in finance and insurance accounted for 4.3% of the nonfarm payroll.] And, for each percentage point above that share, the estimates here suggest a drop in trend productivity growth of as much as 0.4 percentage points. Given that productivity growth in the advanced world is rarely above 2%, that’s a pretty big deal!
Financial sector share in employment and real output growth per worker
Source: Stephen G. Cecchetti and Enisse Kharroubi, “Reassessing the impact of finance on growth,’ BIS Working Paper No. 381, July 2012.
This pattern is consistent with what economists have found when studying the relationship between credit and growth. While there is considerable debate about the significance of the findings, the common result is that when credit – household, corporate or government – exceeds some threshold (measured as a share of GDP), it is associated with slower, rather than faster, economic growth. (See, for example, the papers here and here.) Similarly, recent research on financial efficiency – such as the finding that the unit cost of U.S. intermediation has increased over the past 30 years and that U.S. financiers were significantly overpaid – also supports this view.
The second part of the story is about the growth rate of finance. Here, the evidence appears stronger and more damaging than it is for the scale of finance: as you can see in the chart below, the relationship between the growth in financial sector employment and the growth of productivity is unambiguously negative. The faster the financial industry grows, the worse it is for productivity. (Again, we remove systematic country differences from the chart.)
To get a sense of the how important this effect is, we can look at the case of Spain. During the five years from 2005 to 2010, the share of Spanish employment in financial intermediation grew at the rate of 1.9 percent a year. A decade earlier, from 1995 to 2000, the financial sector employment share was falling at a rate of 3.2 percent per year. Over the period between these two episodes, Spanish productivity growth (measured as real GDP per person employed) rose from zero to +1.3% per year. Estimates based on the chart below suggest that, if the growth rate in the sectoral share of employment in finance had been unchanged rather than rising by 5.1 percentage points over this decade, then productivity growth in Spain would have been a full percentage point higher.
Growth in financial sector employment share and real growth
Source: Stephen G. Cecchetti and Enisse Kharroubi, “Why does financial sector growth crowd out real economic growth?” BIS Working Paper No. 490, February 2015.
Where do these disturbing results come from? Why is it that finance appears to be a two-edged sword, propelling growth in some circumstances, but holding it back in others?
Well, it turns out that the most prevalent forms of finance do at least two things that are not so great. One has to do with collateral and the other with talent. Starting with the first, a growing financial system usually means more credit, rather than more equity. It is difficult if not impossible to get a loan without some sort of guarantee of repayment – a guarantee that is much easier to make if the loan is backed by a building or a machine that the bank can resell in the event of a default. So, businesses with more tangible capital are more likely to get loans. You might call this (as some economists do) the “tyranny of collateral.” Unfortunately, at least in recent decades, these collateral-rich enterprises also turned out to be firms in sectors with low productivity growth.
To see the point, consider that the highest growth sectors tend to be technology and research-based. Computer software and pharmaceuticals are the prime examples. These industries are laden with intellectual property – extremely valuable, but nearly impossible to pledge as collateral for a loan. Compare that to construction. Granted, construction companies do innovate, improving their processes. But, in the end, there is no way that growth in output per worker in construction can compare with that in information technology.
A second reason finance can be bad for growth is that it drains skilled labor from the rest of the economy. This brings us back to where we started. We would probably be better off if at least some hedge fund gurus spent their time figuring out how to produce clean, renewable and cheap energy. Even within the world of finance, we would probably be better off if more of these experts moved into venture capital – where the “tyranny of collateral” does not prevent the allocation of capital to promising (if risky) new opportunities – and away from using leverage and big balance sheets to arbitrage away small gaps in the secondary market pricing of securities.
We can think of additional reasons why finance causes problems for economic growth, and we’ve written about them extensively on this blog (see, for example, our pieces on credit ratings, conflict of interest, Ponzi schemes, and leverage). Most important, when intermediaries fail to perform the critical screening and monitoring necessary to ensure efficient allocation of resources, productivity suffers and the financial system as a whole can become vulnerable to crises.
So, what should we conclude? Many observers (especially the largest financial intermediaries themselves) are critical of the increased regulation of the financial sector following the crisis of 2007-2009, arguing that it will slow economic growth. We, too, can think of some misdirected regulatory efforts that may diminish long-run efficiency without reducing systemic risk (see, for example, our take on raising the maximum loan-to-value ratio for mortgages to 97%).
But, we strongly support the authorities’ efforts to raise capital requirements in order to make the financial system safer. If anything, the research on finance and economic growth strengthens that view, suggesting that there will be little, if any, cost in terms of economic growth even if further increases in capital requirements were to lead to some shrinkage of the size of the financial sector in the advanced economies.
Technical appendix (added March 10): Some thoughtful observers have raised the important question of causality: is strong growth of finance the cause or the result of a slowdown in productivity? One cannot rule out the latter, but there are at least three reasons to conclude that finance is the cause.
First. the papers to which we refer address this question by noting that financial sector growth comes hand-in-hand with credit growth, which is very procyclical. Put differently, it is unusual for the economy (or productivity) to perform poorly when credit growth is high. Second, the argument for reverse causality typically is made in the case of household credit growth as a consequence of poor wage growth. However, the results in the referred papers apply to corporate credit as much as they do to household credit. Finally, the regression results that underlie this blog post are based on an instrumental variables (IV) approach that is designed to address these causality issues. In particular, the regressions use predetermined variables as instruments for financial sector growth. These IV results are nearly identical to those obtained using ordinary least squares (OLS) regressions. Consequently, it seems reasonable to conclude that strong financial sector growth imposes a drag on the real economy, rather than that a weak real economy boosts the growth of the financial sector.