75
5.3.4 Change on total external finance
The total % ∆ in external finance in year T is calculated in the following way:
76 changes significantly over the years. Thus, it is valid to claim all significant changes in Regression I, i.e. level, to be caused by revaluation. If so, changes in level approach the measure return seen from the investor’s point of view. From the NFC’s point of view revaluation first of all have impact on credit risk and the company’s ability to collect more finance.
In short, the countries experiencing the largest decrease in level in 2008 were all bank-based and they do not rebound in 2009 to the same extent as the market-based countries.
Changes in total external finance through time
The ranking in 2007 and 2008 shows a certain consistency in which countries that enjoyed the good times the most are the countries affected mostly by the crisis. However, using a broader time frame (back to 2005), the column Standard deviation show no consistency between volatility in return/revaluation and the financial system characteristic. In the meantime, a pattern in volatility does exist, but as all economies hold three types of finance, the total volatility is a product of:
1. The weights of each type of finance
2. The standard deviation in the marginal contribution factor of each type of finance 3. The correlation between the marginal contribution factors of each type of finance Especially the correlation between each type of finance makes direct interpretation as to why one country is more volatile than another, hard. The marginal contribution factors do also differ from country to country and therefore it is not only the weights and correlation that explain why one system is more affected than another. Looking at the marginal contribution factors individually, as we did above, is therefore much more revealing.
This paper is mostly interested in the financial shock and its impact of NFC financing in the period of 2007 to 2009. This period confirms that bank-based countries were more exposed to macroeconomic slowdown in 2008 and that their volatility from 2005 to 2009 was also the highest.
Risk and return
Interestingly one could already see clear signs of economic slowdown in 2006 for all countries: With no exception, the trend in level, i.e. the marginal change, decreased every year from 2006 to 2008. Only in 2008 the marginal change was negative and so the level decreased. In 2009 the trend in level increased, turned positive, and thus the level increased again. In this sense they all follow the same path.
77 The marginal contribution factors for all types of finance approach each other from 2005 to 2008. In the same manner the % ∆ in total finance also approach each other until 2008. For this to happen, the countries experiencing the highest % ∆ in total finance in 2005 must have experienced the largest change % point drop in the total change. E.g. Norway’s total level increased by 23.08 % in 2006 and US’s 8.79 %. In 2007 Norway’s level increased by 13.47 % and US’s 5.27 %. Norway’s change in level, i.e. trend, from 2006 to 2007 dropped with 9.61 % point or 41.6 %. US’s trend in level dropped with 3.52 % point or 40.02%. Therefore, one could argue that the trend, i.e. change in level, changes with more or less the same for the two countries. However, investors and the NFCs both strive for growth and usually a constant growth. Thus if we put the growth (the change in level) in focus and compare the overall return if the crisis had not appeared with the trend now that it did appear, we see how Norway missed out on bigger return than the US:
Table 5.15 – Example of unrealized return
Country % change in TL 2006
% change in TL from 2006 to 2008
(Assuming the same trend in 2007
as in 2006)
Actual % change
in TL 2007
Actual % change in TL from primo 2006
to ultimo 2007
% point difference between expected return and actual
return
Norway 23.08 % 23.08 % * 1.2308
= 28.41 % 13.47 % 23.08% * (1-.1347)
= 26.18 %
28.41 % - 26.18 % = 2.22 % point unrealized return
US 8.79 % 8.79 % * 1.0879
= 9.56 % 5.27 % 8.79 % * (1-.0527)
= 9.25 %
9.56 % – 9.25 % = -0.31 % point unrealized return Source: Table 5.14
The table provides an example of how much the total level in Norway and the US increased less, than if the return in 2008 had stayed on the same level as in 2007. Of course, that would not have been the case even if the crisis had not become a reality, but it illustrates why countries with the highest return primo the crisis, also missed out on most profit during the crisis. The point can be generalized in the sense that there is a relationship between high profit and high risk of return and this data simply confirm such relation.
78 Statistical caution
Norway missed out on 2.22 % point and US missed out on -0.31 % point level increase. Still, it is crucial to emphasise that these calculations only use the coefficients provided in Regression I and they seek to explain level changes in a specific year as a linear function of the capital-ratio. Table 5.16 provides simple standard deviation from the mean value- figures for the bond level, equity level, and bank loan level from 2003 until 2009.
First, there is no distinct pattern in which type of finance that deviates most from its average level. Thus, for Denmark equity clearly deviated more, but for England bank loan deviated more. Secondly, this stands in contrast to the standard deviation in the marginal contribution factors, which do not take the country specific changes into account. The country dummies serve the role of capturing such difference in level changes and therefore volatility in the marginal contribution factors are not affected and neither is the volatility of total external finance. This is crucial to emphasise. Table 5.16 and the volatility in Graph 5.1, 5.2 and 5.3 in chapter 5.1highligth the total change in level for each country was affected by more and other factors than the capital structure.
Table 5.16 Percentage standard deviation from the average level of external finance
Bonds Equity Bank loans
Norway 27% 39% 37%
DK 14% 41% 31%
Spain 31% 22% 26%
England 20% 14% 24%
France 9% 22% 12%
Germany 9% 13% 6%
US 17% 21% 28%
Finland 10% 25% 20%
Measure: First the standard deviation is calculated from 2003-2009. Then the average in the same period is calculated.
Values in Table 5.16 = Standard deviation / average level.
Source: Appendix 2, real 2003$ values of bond, equity and bank loan level.
Explanation of the change in total external finance through 2009
That the bank-based countries in general were the most exposed ones during the crisis is, as already mentioned as a sum of weight, marginal contribution factors and correlation among these. It is extremely difficult to give a precise answer to why we see this ranking and why the exact % change in level for each country at each point in time turns out as it does. Every single value holds a different underlying story and this story change in any point it time. It is the correlation between the three types of finance that complicates the picture. A thorough
79 analysis will itself require a whole paper, so I will limit the analysis to 2009 to give the reader an idea about the complexity of the % change in total external finance.
The countries that experienced the largest drop in 2008 were not the countries that had the largest increase in 2009. Thus, we cannot claim that we are witnesses to a normal distribution of risk and return based on such a short period of time. The general reason for why the market-based countries experienced the largest increase in 2009 goes as follow: The NFCs in market-based countries are according to the marginal contribution factors for 2009 on equity expected to increase their equity level less than bank-based countries. As E/TL on the other hand weights more for the market-based countries equity’s impact on changes in the total level is still ambiguous. The exact same tendency is true for bonds, however the increase is smaller and so is its share out of total external finance for all countries. The bank loan level decreased in 2009, but very little (1-2 % decrease according to Table 5.12). The decrease in level reduced the change in total external finance for all countries, and the reduction should weigh more for bank-based countries as they hold a larger bank loan - ratio. Thus, the total explanation for why market-based countries experienced a larger increase in value in 2009 is first of all a consequence of bank-based countries experiencing a heavier weighted decrease in bank-loans.
According to the marginal contribution factors we would see a higher volatility in the total level for bank-based countries throughout the whole period. Thus, if it was only the marginal contribution factors that determined the percentage change in total external finance we should have seen a larger increase for bank-based countries in 2009. That this is not the case is simply due to the difference in the weight of bonds, equity and bank loans. During the crisis market-based countries simply outperformed the bank-based countries due to the mix of external finance and not because the single type of finance increased more in 2009 than the corresponding financial sources in bank-based countries.
80
6 Discussion
This chapter analyzes and discusses the relationship between the results (chapter 5) on one side and the seven hypotheses (chapter 3) and theoretical content (chapter 2.1 and 2.2) on the other. Macroeconomic data is introduced to support causality between the empirical findings and the hypotheses. Many hypotheses treat both research question A, B and C.
In order to make valid conclusions causality must be proved between the results and the reactions (and reasons for such reactions) to the crisis as postulated in the hypotheses. Such proofs require their own examination to an extent that goes beyond this paper, if they are to be valid. Thus, any causality presented now are merely indicators of what to expect if we decided to conduct new and more thorough calculations on each hypothesis. However, the results presented in chapter 5 are indeed valid. Still holding these results against new macro-economic data merely indicate correlation and causality. All macro-macro-economic data is quoted in current prices.
Most macro-economic data quoted in current prices focuses on the balance sheet of banks; their allocation of assets and credit tightening. Banks, like everyone else, address the return on assets and credit tightening in the currency and price used at the day where the go to work and this is the quotation they react to. They to not say: “Measured in 2003 $ prices the credit tightening is actually not so bad”. I believe banks to react of what happened today, measured in current prices and it is their reaction to the crisis that is interesting.
For balance sheet information England is not included. The reason is that England does not use the same balance sheet accounts and thus the data is incomparable. However, as the discussion chapter do not claim to be perfectly statistical valid, but merely to show indications, it is acceptable that England is not included. However, it would have been preferred.
Hypothesis I
Effect 1, 2 and 3 argue for and against banks’ enhanced capital access. Summarizing effect 1 and 2, I would argue that depositors’ lack of confidence in banks is larger than their lack of confidence in the secondary market, thus the combined effect is in favour of bank-loans.
Effect 3 on interbank lending on the other hand is expected to have significant impact on total liquidity and liquidity risk. In order to enhance liquidity and thus reduce the liquidity risk, all types of assets are expected to be reduced. This will mostly affect bank loans, unless security holders others than banks reduce their holding of securities with the same degree as banks do.
81 Bank loans are the least liquid type of financial assets, so they are reduced more slowly than securities. The total effect of reduced interbank lending is expected to countervail higher consumer deposits. The reduction in the NFC’s securities are at first more severe, but reduction in bank loans will continue over a longer time.
Discussion of hypothesis I
Before looking into how the total level and allocation of bank assets changed during the crisis, the first step is to prove whether or not customer deposits and interbank deposits changed. If none of these changed, there is no point in looking further into the changes in bank assets. The focus is on the period of 2007-2008, because the interest is on the first period of the crisis.
Changes in bank deposits
Appendix 11.1 shows the total level of customer deposits and interbank deposits. For consumer deposits, the level increased through 2008, but only the same or less than it did in 2007. Contrary to expectations, there is no evidence indicating a higher increase in preference for deposits than in the previous periods. For interbank deposits only the level in Norway and Finland increased through 2008. For all other countries the interbank deposit stagnated or decreased and is therefore more affected by the crisis than consumer deposits. Exactly as expected. The point is even clearer when compared to an otherwise constant increased in interbank deposits since 2005 until 2008. Combining the effect on consumer deposits and interbank deposits, we therefore see that the banks’ access to short-term capital increased less than it did in the previous years. We do only see an actual decrease for Denmark and maybe Germany. Thus, to the extent that a decrease in the trend has any impact on the asset level and allocation of banks, the impact is expected to be moderate.
Changes in the level of bank assets
Banks investment in securities, i.e. equity and bonds, follows a rather unstable increase each year until 2008. In 2008 a more systematic slowdown in a still positive trend appeared. For France and Germany value of securities even dropped (Appendix 11.1) 14. The level for bank loan assets stands out: It increased with an increasing trend for all countries through 2004 and 2005. In 2006 and 2007 the increase in level stagnated (the level increased with a constant trend) and in 2008 the increase in level slowed down to around 7% to -3 % (depending on the country). Thus, the slowdown (measured by the change in level, not the total level) began in
14 OECD does not provide bank asset data for 2009 yet.
82 2006, i.e. before any change in deposits. The change in bank loan assets approached each other for all countries, i.e. the change was more systematic than before. Thus the reduction in level trend from 2007 to 2008 was largest for the countries enjoying the highest increase in 2007. These countries are Denmark, Spain and Norway.
Causality between changes in deposits and changes in external finance for NFCs
So, we see a positive correlation between the combined effect of changes in the customer and interbank deposits and changes in security and bank loan levels.
Since the signs of slowdown in bank loans happened already in 2006, other factors must have played a role than changes in deposits. This is not a claim, that a decrease in the trend of deposits does not impact banks investment in bank loans. However, the impact is not convincing. If we accept that banks reduced their holding (or slowed down their increase in holdings) of bank loans, it can happen through (i) reducing in new lending and (ii) redemption of loans. Both actions should be directly and fully observable on the NFCs net-transactions.
Two exceptions exist: If the reduction of loans affects foreign debtors more than domestic debtors and if the reduction affects other debtors than NFCs, e.g. private debtors. We know that net-transactions of NFCs did not change significantly during the crisis and thus we can reject that banks changed their lending policy to NFCs - no matter the trend in deposits.
Change in the trend of bank loans held by banks must then have been explained by either reductions of loans to foreigners, privates or revaluation. None of these are of any relevance to hypotheses I.
Reduction of banks’ investment level in securities can happen by net-selling or revaluation. Net-selling is not necessarily observable on the NFCs’ net-transactions in securities, because securities can be sold on the secondary market. Only trading where the NFCs are directly involved is registered as net-transactions. Revaluation of existing securities puts banks in a passive position where they receive the market price but keep their quantity constant. With the data at hand it is not possible to tell what effect that caused a slowdown in security holdings. If it was the market value that changed, we would be able to see the effect on NFCs’ level of bonds and equity. However, since net-selling of securities held by banks all things equal also reduces the market price, it is not possible to separate the effect of value reductions due to other market players and value reductions due to active net-selling by banks. Thus I cannot conclude if NFCs partly experienced a reduction in their securities due to an active policy by the banks to increase their liquidity.
83 Further research
In reality, the reaction of banks to the crisis probably happens in parallel to even the smallest signs of potential recessions. These reactions are exactly what stimulate further value reductions in the financial market. To assess whether lower increases or stagnation in deposits motivated banks to sell securities and whether this affected the value of NFCs securities would require monthly, preferable daily, observations where lagged values of the % ∆ in deposits are used as explanatory variables and held against the % ∆ in security values as stated on the asset side of banks balance sheets. The changes in security values would furthermore need to be separated into changes due to net-transactions and changes due to revaluation. The task is beyond the scope of this paper.
Hypothesis II
Effect 4 states that government bonds are expected to serve as an appealing alternative to deposit savings thus lowering the risk free interest rate premium, the effective bond rate and equity returns. The impact on equity is expected to be limited, as the interest rate premium accounts for a small fraction of the total cost of capital. Effect 12 adds the effect of expansionary monetary policy. This results in a higher demand of bonds which lowers the interest rate premium even more. The total effect is expected to be significant simply because the agenda of central banks was to stimulate investments and consumption. However, the effect is expected to be greatest for bond levels, then bank loans levels and the least affected capital source is expected to be equity. The effect is not expected until the central banks assess that the risk of real economic slowdown is present.
Discussion of hypothesis II
In order to address hypothesis II correctly, it first must be concluded if consumers changed their preferences for saving in banks during 2008 and 2009. If consumers changed preferences towards government bonds it should be reflected in the yield of these. The yield is measured as short-term interest rate on T-bills. Thus, this is the second point of interest. Then, potential discrepancy between movements in the yield of government bonds and saving deposits is discussed. In the end it is discussed what the consequences of changes in the monetary policy are for external finance of NFCs.
Consumer deposits
Graph 6.1 shows that the trend decreased to around 5 % during 2008 and almost stagnated through 2009. Not since 2003-2004 had the increase in deposits been so low. Thus customers
84 slowed down their increase in bank saving for one reason or another, but the deposits still increased with 7 %- 15 % for all other countries than Germany and Denmark.
Graph 6.1 – Yearly change in consumer deposits
Graph 6.1 use aggregated balance sheet information for all types of banks. Deposits are stated as a liability as quoted in current prices in millions of national currency. % Changes in deposits = (Dept-Dept.1)/Dept-1. More observations would have been preferred. Source: OECD financial statistics, also available in Appendix 12.
Change in the short-term interest rate
The short-term interest rate is derived from interbank lending rates and rates on Treasury bills. Banks lend via the central bank and the central bank issues bonds or bills to the public.
The central bank charges a very little margin on top of its own interest expense and therefore these two rates are used indiscriminately. T-bills are, due to their low risk profile and short time to maturity good alternatives to storing money in saving accounts.
Graph 6.2 – Short-term interest rate on government bonds
Graph 6.2 shows the yield on short-term interest rates, quoted monthly. It is derived from a mix of 3-month inter-bank rates or T-bills. Source: OECD Financial Indicators (MEI).
-3%
2%
7%
12%
17%
22%
27%
% ∆ in customer deposits
% ∆ in consumer deposits
Denm ark Finlan d France Germa ny Norwa y Spain
0 1 2 3 4 5 6 7 8
% Yield per annum
Short-term interest rates
Finland France
United Kingdom United States Denmark Spain Germany