One piece of conventional wisdom, often (only?) touted in the financial industry, is that the bigger the term premium the bigger the bank profits. This relates to the simple observation that one of the most basic tasks of banks is maturity transformation, i.e., the bank borrows short-term and lends long-term, and pockets (sort of) the difference. This is sustainable because…banks are experts and do this wisely. And hold a basket of different eggs (loans) that they attach exactly the right premia to. And…– Yep, that’s how it works!
Only in America!
In this post I will discuss and dissect work by the Bank of England and the NY Federal Reserve, and others, that have quite recently been released. As usual, the best data freely available pertains to the US economy, and we see a declining trend in the net interest margin (NIM) over the last 25 years or so. The low NIM period preceding the historical peak in the below chart is probably, at least to some extent, a result of the Savings & Loans crisis engulfing the US financial sector in the late 80’s and early 90’s.
Chart 1: The US Net Interest Margin
The most recent downturn in NIM seems to have reached its nadir in 2015, coinciding with a peak expansion in US monetary conditions as we identified in last week’s post. This might add some color to the financial market maxim related to the above, but what is really the net effect of both higher interest incomes and expenditures? Does it imply an improvement of bank profits in the period ahead, at the same time as the FED raises rates and tapers its balance sheet? The NY FED-post cited above does quite a bit of this, posting a regression where the 10y-3m term spread is set to explain the net interest margin.
Chart 2: Slope is intact. Constant is not.
In the graph above I have tried to expand a bit on the NY FED post because, for starters, the original text does not show that while the slope coefficient has the expected positive sign, the test statistics are very poor (when looking at the full sample). It is only when you divide the sample into sub-periods that any patterns emerge1.
The slope coefficient is quite (well) stable over time but increases gradually from ca +0.05 in the first period to ca +0.15 in the latter period), suggesting that the term spread is becoming increasingly important for bank profits. And whereas the slope coefficient seems stable, the constant is anything but; it falls gradually throughout the sample. How do we interpret a decreasing constant? – Well, many probably see it as decreasing credit premia, which could in turn be interpreted as a sign of increasing competition, something that may sound counterintuitive as the banking industry is considered a hallmark of oligopolistic power.
Is it possible that the banking market is becoming more of a “contestable market” as fin-tech and international competitors are making inroads, diminishing some of the protection from sunk costs and barriers to entry/exit? – If so, it is somewhat ironic that over the past years regulators have been doing their best to raise these protections by increasing sunk costs and barriers to entry. Simultaneously, however, regulators are also decreasing the possibilities of profit from maturity transformation, requiring higher maturity matching of banks’ assets and liabilities. On this point, if anything, our simple example points to the opposite, with term premia becoming more, not less, important.
As a final note on the US experiences, if the constant continues to shrink from competition (or whatever) at the same pace as over the past 25-30 years, it means that in approximately another 25-30 years the NIM will be almost completely governed by term premia (and closer to a perfectly competitive market).
1Note that we can still only explain between ¼ and ½ of the variation.
International and individual bank experiences are different
Next, let us look at international experiences. We can do this in a number of ways, but I have chosen to add together the Net Interest Income for individual banks from the Equity Fundamentals node in Macrobond and differentiate these series (it is possible to relate it to various asset measures as well). The calculations make it possible for you to, e.g., look at individual banks and, of course, improve on my calculations. For comparative purposes, I have also run the calculations on data from the World Bank database (available in MB) which includes, e.g., information on Net Interest Margins (alas, only on annual basis) for quite a few countries and regions.
Under any circumstances, comparing the results to the Bank of England research (also mentioned above) is quite telling. Whereas the BoE finds a positive slope only for the US (negative for all other economies), our results are considerably more ambiguous and do show a positive slope for most countries and aggregates.
Note: Mind the scale on the axis. In the attached MB-documents you can find individual banks and other country aggregates.
Charts 3a-d: After the GFC, the maxim holds
Granted, this could be due to differences in data coverage, but comparing between data sets and changing time windows it seems, as in the US case above, to mainly be a result of how far back the data goes. The ‘older’ the information we include, the more often our result suggest a negative slope. A negative slope, to be sure, implies that a higher term premium is associated with reduced net interest margins – the opposite of our financial market maxim above.
Under any circumstances, please note the wide dispersion, and consider effects of outliers. E.g., for the US calculations, we would see a considerably higher slope coefficient if outliers were excluded. Nonetheless, at least the US results rhyme with our previous findings, and also those of both NY FED and BoE. However, when looking at UK and other country aggregates you see that the BoE-result of omnipresent negative slopes don’t hold in the calculations. There are probably a number of reasons for this; I have already mentioned coverage and the time span, BoE estimates their regression on data going back to the 1980’s or even earlier. But to be honest, we (and in particular BoE) should perhaps also control for other variables since the chosen time period is one of great structural change(s) in the banking sector. Moreover, the wide dispersion makes any conclusions very difficult, a complication which neither BoE nor NY FED comments on. This is, not least, illustrated by the very poor test statistics in most of our calculations. That said, the thrust of our results go hand in hand with those of Claessens et al. (2017), who use a time window similar to ours (but add some valid controls).
Even though the relationship between the term premia and banks’ net interest margins seem to be vaguely positive, as expected, our simple calculations suggest that in and by itself, term premia are not a very good indicator of net interest margins. Albeit a good starting point, recent posts on this issue from NY FED and BoE should not be taken at face value as it seems other factors, both cyclical and structural, are having a noticeable impact on the NIMs. Even when we look at individual banks or well-defined banking systems, such as the Nordics, the relationships are still quite haphazard.
Charts 4a-d: You can’t hide the wide dispersion
Here, again, we discern some type of structural change post-GFC, as the constant decreases and the slope coefficients seem to increase (not in Norway though) implying a starker effect of the term premia. Moreover, it seems fitting that Denmark and Finland have a stronger inclination as longer, fixed interest rate loans are more common there than in Sweden and Norway. But the overall impression when looking at the distributions remain: It is hard to discern any clear pattern.
So, what are we left with? – Well, the financial market maxim of ‘the bigger the term premia, the bigger the banks’ profits may still hold. We simply cannot say based on our (nor from the NY FED or BoE) simple calculations. Instead, what is suggested is further exploration where controls for a number of both cyclical and structural factors are added.
– Indeed, that is perhaps the most intriguing result of our calculations; no matter what level of aggregation or which set of data I use, it seems as if a decreasing intercept is the…one thing constant (!). For how long can it go on? What does it depend on? Now, that’s an issue for our clients (and Bank CEOs) to fret about!
Disclaimer: We don’t usually have views and opinions about economic and financial states of affairs, (not ones that we express publicly as a company, anyway). We do believe, however, that people can and do appreciate a variety of perspectives. What you’ve just read is the perspective of the author. While we think our writers are very smart, Macrobond Financial does not expressly endorse the views presented here. And, as the old adage goes, you shouldn’t believe everything you read (not without finding the data, performing a few analyses and presenting it in a nice chart). We want to make it clear that we are not offering this information as investment advice. That being said, if you have Macrobond, you can easily check everything that’s mentioned here, and decide for yourself. If you don’t have Macrobond, now you have a great reason to get it.