- Email: [email protected]
A financial accelerator in the business sector of a macroeconometric model of a small open economy
Journal Pre-proof A financial accelerator in the business sector of a macroeconometric model of a small open economy Andreas Benedictow, Roger Hammersland
PII:
S0939-3625(18)30057-8
DOI:
https://doi.org/10.1016/j.ecosys.2019.100731
Reference:
ECOSYS 100731
To appear in:
Economic Systems
Received Date:
10 October 2017
Revised Date:
15 December 2018
Accepted Date:
13 January 2019
Please cite this article as: Benedictow A, Hammersland R, A financial accelerator in the business sector of a macroeconometric model of a small open economy, Economic Systems (2019), doi: https://doi.org/10.1016/j.ecosys.2019.100731
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier.
A financial accelerator in the business sector of a macroeconometric model of a small open economy Andreas Benedictow a and Roger Hammersland b a
Statistics Norway; E-mail address: [email protected]
b
Statistics Norway; E-mail address: [email protected]
Highlights ·A financial accelerator mechanism operating through investments in the business
ro of
sector has been incorporated into a dynamic macroeconometric model of the Norwegian economy
·In this new and amended model, aggregated credit and equity prices are determined
-p
simultaneously in a system characterized by a two-directional contemporaneous causal link, whereby higher equity prices spur more credit and vice versa The sub-system of credit and equity prices has been designed and estimated with a
re
new procedure of simultaneous structural model design
Combined with a mechanism where credit and asset prices are mutually influenced by
credit-asset price spiral.
Simulations illustrate how the introduction of a financial accelerator significantly
na
lP
real investments in the short run, this creates a financial accelerator amplified by a
reinforces and extends the economic cycles in projections and forecasts.
A permanent increase in global equity prices of 10 percent is calculated to increase
ur
Norwegian asset prices by 10 percent in the short run, gradually decreasing to 5 percent in the long run. As credit and equity prices are only allowed to affect the short-term dynamics of
Jo
capital accumulation, this leads only to a minor short-term increase in real business investments, which gradually peters out in the long run.
Monetary policy gets a markedly stronger effect in the short and medium term, while the impact of fiscal policy is affected to a relatively small degree as it is more remotely linked to financial markets
Abstract 1
We have incorporated a financial accelerator mechanism operating through investments in the business sector in a dynamic macroeconometric model of the Norwegian economy. In this new and amended model aggregated credit and equity prices are determined simultaneously in a system characterized by a two-directional contemporaneous causal link, which has been designed and estimated by a new procedure for simultaneous structural model design. Combined with a mechanism where credit and asset prices are mutually influenced by real investments, this creates a financial accelerator amplified by a credit-asset price spiral. Simulations illustrate how the introduction of a financial accelerator significantly reinforces and extends the economic cycles in projections and forecasts, in particular when confronted
ro of
by a severe shock. Furthermore, monetary policy has a markedly stronger effect in the short and medium term, while the impact of fiscal policy is affected to a relatively small degree as it is more remotely linked to financial markets.
JEL classifications: E1, E32, E44
-p
Keywords: Financial variables and the real economy, The financial accelerator, Business
re
cycles, Structural vector error correction modelling, Impulse response analysis, Forecasting
lP
1. Introduction
The idea that conditions in credit markets could affect business cycles has had broad support in the economic literature for many years; see, for example, Hubbard (1998) and
na
Bernanke et al. (1999). The theory of a financial accelerator postulates a reciprocal relationship between access to credit and fixed investment that helps amplify cyclical
ur
fluctuations, see Bernanke and Gertler (1989). Kiyotaki and Moore (1997) took this further by
Jo
introducing an explicit equity price and credit spiral. There has emerged a substantial empirical literature that largely found support for a relationship between (various indicators of) credit availability and macroeconomic fluctuations, see for example Silvestrini and Zaghini (2015). These works were based largely on the equilibrium models of the real business cycle literature (RBC) (see Kydland and Prescott, 1982; Hartley, 1998). In addition, to some extent there are implemented financial accelerator mechanisms in the so-called new
2
Keynesian DSGE models (see Smets and Wouters, 2008; Christensen and Dib, 2008). However, few attempts have been made to incorporate such a mechanism into structural macroeconometric models. An exception is Hammersland and Træe (2014), where two reciprocal and interacting financial accelerator mechanisms are implemented in a macroeconometric model (Bårdsen and Nymoen, 2009) to study the effect of different types of shocks on the financial stability of the Norwegian economy. This model is, however, highly aggregated, and although it contains financial accelerators with origins in both the
ro of
household and business sectors, the interaction between the real economy and the financial variables happens directly via aggregate production (GDP Mainland Norway) and not, as according to economic theory, via its structural sub-components, household consumption and
-p
business investment, respectively.
This paper documents the estimation and implementation of a financial sub-model in
re
a structural macroeconometric model for the Norwegian economy, KVARTS, partly inspired
lP
by Hammersland and Træe (2014).1 However, our implementation is more disaggregated and theory consistent than in previous studies, 1) in that the financial variables affect investment directly and 2) by taking into account that the effect of changing credit and equity prices on
na
investments can be industry-specific. KVARTS is expanded by a financial sub-model where aggregate credit and equity prices in Norway are determined simultaneously in a system
ur
characterized by a two-directional contemporaneous causal link, designed and estimated with
Jo
the help of a new procedure for simultaneous structural model design (Hammersland, 2017). Moreover, the equations in KVARTS for capital formation in each industry are expanded with aggregate credit and/or Norwegian equity prices. The industry-specific real capital is then aggregated to total capital formation, in which the change from one point in time to the next is 1
KVARTS is developed by the research department of Statistics Norway. A full description of KVARTS is beyond the scope of this paper. See Boug et al. (2013A, Appendix A) for an outline of KVARTS. We also refer to Bowitz and Cappelen (2001), Boug et al. (2006), Boug and Fagereng (2010), Benedictow and Boug (2012), Jansen (2013) and Boug et al. (2013A, 2013B) and Hungnes (2016) for descriptions of the main sub-sections of the model.
3
defined as investment in the same period and included as an explanatory variable in the financial sub-model. Thus, equity prices and credit in this model affect real investments, which in turn affect equity prices and credit, and so on. In the financial sub-model, aggregate credit to the mainland industry in the long term is determined by Norwegian equity prices, represented by the Oslo Børs benchmark index and aggregate business investment in mainland Norway, in the short term also by real interest rates, all deflated by the deflator for GDP Mainland Norway. Norwegian equity prices
ro of
are in the long term determined by international equity prices, represented by the global equity price index Morgan Stanley Capital International World (MSCI), real oil prices and the real interest rate, in the short term also by credit to non-financial corporations in mainland
-p
Norway (henceforth referred to as credit).
In KVARTS, gross fixed capital formation (JK) is divided into two main groups of
re
industries; 1) investments in extraction and pipeline transport, and 2) investment in mainland
lP
Norway.2 Investments in the latter group can be divided further into a) investments in public administration, which is an exogenous variable in the model, b) housing investment, which is determined in a separate sub-model for the housing market, where there is also an accelerator
na
mechanism between credit and investment (see Jansen and Anundsen, 2013), and c) business investments, which is the group of industries directly affected by the financial accelerator
ur
presented in this paper.
Jo
The capital stock is determined in KVARTS by 13 estimated, industry-specific equations. The explanatory variables are production and relative factor prices and other relevant variables, such as employment.3 Norwegian equity prices and credit are only
2
Strictly speaking we could also include investments in shipping as a third group here. However, these investments amounted to just 1 percent of the total gross fixed capital formation in 2014. By comparison, 1) was about 29 percent, while 2) a, b and c represented 30, 21 and 20 percent, respectively. 3 Employment (hours worked) is determined in a factor demand system in the same way as real capital, by production and relative factor prices, and, additionally, technological progress (represented by a deterministic trend).
4
included in the short-term dynamics of each equation, and consequently do not affect capital stock in the long term. This is in line with the Modigliani-Miller theorem (Modigliani and Miller, 1958). We find support for an effect of Norwegian equity prices and/or credit in all equations. Gross investment in each industry appears by definition as the change in capital stock from the period before, adjusted for depreciation. The importance of the financial accelerator for the economy as represented in KVARTS is illustrated by three exogenous shifts and a counterfactual experiment that is
ro of
based on a fairly recent experience related to the Norwegian economy. We start with a shift in the global equity price index, MSCI, which only appears in the financial sub-model. A permanent increase of 10 percent in the MSCI leads to a rise in Norwegian equity prices of 10
-p
percent during the first two quarters, followed by a rapid decline gradually decreasing in strength, and converging towards a long-term effect of approximately 5 percent after 6-7
re
years. As credit and (Norwegian) equity prices are only included in the short-term dynamics
lP
of the capital equations, we only get a short-term increase in business investment at just over 1 percent, which gradually disappears within ten years. We then show the additional effect of changes in the money market rate and public demand, respectively, which is attributable to
na
the financial accelerator in KVARTS. A permanent reduction in the three-month money market rate of 1 percentage point leads to a short-term additional increase in business
ur
investment of 1.4 percent after three years due to the financial accelerator. After 8-10 years
Jo
this effect is gone. Next, a permanent increase in government consumption, investment and employment by 1 percent provides only a marginal additional increase in business investment. The additional effect is weak because there is no direct link in the model from public spending to business investment and financial markets. Finally, we look at a counterfactual shock to oil prices and oil investments where these quantities, instead of
5
following their historical paths4 given by the baseline scenario, are set to follow a random walk, basically implying that we maintain the level of these variables at the beginning of 2013 over the entire simulation period. As the data of our baseline scenario implies a substantial and protracted fall in oil prices as well as oil investments of close to, respectively, 40 and 30 percent at the end of the historical sample period in 2017 and only a gradual increase thereafter, this would amount to a substantial boost to the Norwegian economy. If we look at the additional effects of such a counterfactual shift that are due to the inclusion of a financial
ro of
variable per se, these clearly turn out to bear out the relative importance of the financial accelerator in the wake of shocks, as the additional effect attributed to such a model extension is simulated to give an additional boost of close to 1 percent to GDP.
-p
Our results show that introducing a financial accelerator significantly reinforces and extends the economic cycles in the projections and forecasts in KVARTS. In particular,
re
assumptions about future developments in international equity prices prove important for the
lP
estimated business cycle. Monetary policy gets a markedly stronger effect in the short and medium term, while the impact of fiscal policy is affected to a relatively small degree. As far as the counterfactual experiment is concerned, it clearly contributes to demonstrating the
na
intensive role of the financial accelerator in the propagation of severe shocks. In Section 2, we explain the theoretical background of a financial accelerator and
ur
discuss the estimation results for the financial sub-model. In Section 3 we discuss the
Jo
background for the capital equations in KVARTS, which is the point of connection for the financial sub-model, and the estimation results. In Section 4 we highlight the impact of introducing a financial accelerator into a large macroeconomic model by comparing the effects of shifts in exogenous variables in KVARTS with and without the financial accelerator. Section 5 summarizes and concludes. 4
As this experiment is based on simulating the model from the beginning of 2013 to the end of 2020, the last part of the historic data refers to our quarterly forecasts for the period that goes beyond the period for which we have actual data.
6
2. The financial sub-model 2.1 Procyclicality and the financial accelerator The main point of connection between the real economy and financial markets is the private sectors’ need for the external financing of investments. External financing can be obtained either through an expansion of equity by issuing shares or by increasing foreign capital through borrowing.
ro of
The hypothesis of a financial accelerator assumes that equity prices are procyclical. Increased economic activity will lead to higher equity prices, which in turn can give rise to increased investments, higher production and so on. Such a self-reinforcing mechanism can
-p
be explained by standard economic theory. For example, procyclical behavior in the business sector can be justified by increasing equity prices, causing the market price of capital to
lP
This in turn, increases investments.5
re
increase relative to its replacement cost, the relationship known as Tobin’s Q (Tobin, 1969).
However, what is known as a financial accelerator in the literature is strictly speaking an addition to the classic procyclicality described above and arises from the presence
na
of so-called financial frictions. Financial frictions denote conditions that disrupt the players' behavior in the financial market, and in principle cover all costs associated with financial
ur
transactions, be it fees, taxes, time spent, asymmetric information, etc. (see, for example,
Jo
Brunnermeier et al., 2012, for a literature review). Asymmetric information can for instance cause banks to limit their lending to investors who cannot provide sufficient collateral, so that otherwise profitable investments are not being undertaken. Under such circumstances, previously rationed investors may increase borrowing and realize new investments as higher equity prices boost collateral. This may in turn cause the equity prices to rise further, and so
5
Friedman’s permanent income hypothesis (Friedman, 1957) describes a similar mechanism in the household sector, linked to a positive wealth effect on consumption.
7
on. This illustrates how a financial accelerator with a credit and equity price spiral may reinforce economic cycles. Financial frictions are one possible explanation for deviations from the classical financial theory of market adjustments and the hypothesis of efficient markets launched by Eugene Fama (1965).6 Stiglitz and Weiss (1981) and Stiglitz (1982) argue that asymmetric information in particular poses a significant problem and can explain financial instability as well as financial crises. See Brunnermeier (2001, 2008) for literature surveys on bubbles and
ro of
financial frictions and Jermann and Quadrini (2012) and Hirano and Yanagawa (2016) for more recent studies. The international financial crisis around 2008 intensified the interest in financial frictions and their effects on the real economy. Hall (2009) finds empirical evidence
-p
for the existence of such frictions and the importance for financial markets as well as the real economy. Stiglitz (2010) argues that the effects are significant and that the authorities may
re
advantageously reduce frictions through economic policies. Adrian et. al. (2013) also find
lP
some evidence for financial frictions, although the business sector’s overall access to financing due to the financial crisis was largely maintained because bank lending was to a great extent replaced by bonds. Hammersland and Træe (2014) find clear evidence for
na
financial frictions in the Norwegian economy.
ur
2.2 Empirical results for the financial sub-model
Jo
The financial sub-model in KVARTS consists of two econometric equations where aggregate credit7 and equity prices are determined simultaneously and interactively, and total
6
The field of behavioral finance offers an alternative explanation model, which is not bound by classical economic assumptions about rational actors and efficient markets, but instead has its basis in psychology discipline hypotheses about human behavior. This path is not followed in the present paper (see, for example, Diamond and Vartiainen, 2012, for a discussion). 7 Actual credit, i.e. the market solution: We are not able to distinguish between supply and demand for credit.
8
gross investments in the business sector is included as an explanatory variable.8 This specification will capture the presence of both classical procyclicality and a financial accelerator as described above, but cannot tell them apart. For simplicity, we will refer to all procyclicality arising through the financial sub-model in KVARTS as a “financial accelerator”.9 The financial accelerator is estimated and designed simultaneously with a new procedure for simultaneous structural model design (Hammersland, 2017).10 Based on an
ro of
exact identified general model structure, the final dynamic model is designed and estimated simultaneously using the maximum likelihood method. The first step in this procedure is to identify the long-run solution using the methodology of Johansen (1995). There we found 8
Jo
ur
na
lP
re
-p
A large proportion of non-financial enterprises in mainland Norway is not listed, but the main index on the Oslo Stock Exchange can be an indicator for the development in unlisted companies as well. 9 Two alternative methods were considered for incorporating a financial accelerator in KVARTS. The second was to estimate new equations for investments in every single industry, where the industry-specific investments, aggregate credit and equity prices were estimated simultaneously as a three-dimensional structure. Applying this methodology, however, only two industries (81 and 85) turned out to be suitable for empirical modeling in line with economic theory. It may be that there are conditions in macro that are not captured in the disaggregated figures. Another problem was that we did not have access to industry-specific figures for credit and equity prices. Thus, we concluded that the most appropriate approach is to estimate the financial accelerator at an aggregated level and consequently include aggregate credit and equity when estimating the industry-specific capital equations, as described in this report. 10 To give the reader an idea of what this procedure is all about, we here give a brief account of the steps involved (for a more profound and detailed treatment, the interested reader is referred to Hammersland, 2017). The point of departure in the general case is an n-dimensional conditional reduced form VAR of order k where the conditional set of variables, in addition to including ordinary exogenous variables, deterministic terms and dummies, possibly includes a set of “structural” variables that can later serve the role of auxiliary tools to help with the exact identification of the structural model, both in the long and short run. Starting out with what hopefully constitutes a congruent general unrestricted reduced form model (GUM), the first step then involves reducing the model down to a more parsimonious order and then to undertake the long-run analysis (identification, design and estimation) on this version of the model by resorting to the multidimensional procedure of Johansen (1995). In addition to theory, exogenous and deterministic variables with a structural information content (earlier referred to as “structural” variables) can here be utilized to accomplish exact identification. Given exact identification is accomplished, the next step then involves designing the parsimonious version of the long-run structure by letting theory in conjunction with a test of overidentifying restrictions inform the rest of the long-run design process. Having thus arrived at the long-run structure of the model, one then maps the reduced form of the model over to a structural representation (or form) utilizing some of the “structural” exogenous variables included at the outset as tools of exact identification, possibly in conjunction with restrictions on the long-run feedback structure of the model. It is important in this respect to realize that by utilizing so-called structural exogenous information, or restriction on the long-run feedback structure, we can thus accomplish exact identification without having to resort to procedures imposing non-testable restrictions on either the contemporary feedback matrix or the covariance matrix, both attended with highly contentious and controversial issues given their important role in conditioning the properties of the model. After having designed the long-run structure and exactly identified the dynamic structure, conditional on this, the last and final step of the procedure then involves a fully simultaneous and structural reduction (or design) process where all the structural equations are jointly designed by letting tests of overidentifying restrictions inform the reduction process.
9
support for three cointegrating vectors for credit, equity prices and aggregate investments, respectively. Credit is homogeneous of degree one in equity prices and investment, while the equity price is homogeneous of degree 1 in oil prices and international equity prices, plus an additional interest rate effect. Investment is a function of the relationship between equity prices and the replacement cost of capital, i.e. Tobin’s Q. As investments in KVARTS are already determined by the capital equations, the investment equation in the estimated system is replaced by aggregate investments as determined in KVARTS, i.e. the identity presented in
ro of
equation (7) in Section 3.1. Thus, when we estimate the dynamic simultaneous structure in equation (1), the investment equation is taken out of the system.11 The broad (real) equity price index on the Oslo Stock Exchange (rborsi, hereafter referred to as Norwegian equity
-p
prices) and real credit to non-financial corporations in mainland Norway (rk2nff, hereafter referred to as credit) is precisely identified by assuming that the deviation of credit from its
re
long-term equilibrium, rk2nff-0.77rborsi-0.23jk, where jk is real industry investments, only
lP
helps to explain the structural, dynamic course of credit, while disregarding contemporary effects from changes in international equity prices (msci) in the dynamic relation for real
11
ur
na
credit.12,13,14
Jo
See Appendix B Table B.5 for the econometric specification of the long-run aggregate investment relationship. Confer also footnote 15. 12 All variables are deflated by the GDP deflator for mainland Norway, except MSCI, which is interpreted as an indicator of developments in international financial markets. 13 This is just one of many possible ways of accurately identifying (initializing) the design process. Alternatively, one could for example have chosen to ignore the fact that a discrepancy between one of the endogenous variables and its long-term solution can affect the dynamic course of one of the other endogenous variables in the system. One could also have made use of a priori information on the structural properties of some of the deterministic variables in Dt (structural dummy variables), while imposing alternative restrictions on the system’s Г-matrices, i.e. on the coefficients that capture the effects of exogenous factors and lags of the dependent variable as another option. 14 The long-term structure here is overidentified and represents the result of a long-term analysis where the exact identifiable restrictions imply the absence of oil price/international stock price effects in the credit equation and homogeneity of degree one between real stock price, real oil prices and international equity prices in the longterm solution for equity prices. As in the case of the dynamic structure, this is just one of many ways of accurately identifying the long-term structure.
10
k 1 b12 rk 2nff rk 2nff y , j rborsi t j b21 1 rborsi t j 1
(1)
jk k rpoil x , j Dt msci j 1 RR t j
0 rk 2nff 0.77rborsi 0.23 jk 11 11 11 12 0 22 rborsi 0.5rpoi ln ok 0.5msci 0.04 RR t 1 21 22 23
jk 13 rpoil 33 msci RR t
Other explanatory variables in the system are the real oil price in Norwegian kroner
ro of
(rpoilnokt), the real interest rate (RRt) and deterministic variables (Dt) like a constant, seasonal dummies and dummies for structural breaks. Lower case letters indicate logs and Δ indicates change from the preceding period. Θ, Гx and Гy represent coefficient matrices for the effects
-p
of deterministic variables and dynamics, while b, γ and α represent coefficients that capture effects of the contemporary causal relationship between the model endogenous variables,
re
contemporary dynamic effects of changes in exogenous factors and equilibrium correction.
lP
Because of a general lack of automatic “general to specific” modeling algorithms for structural systems, the structural design and reduction process has been carried out manually on quarterly data from Q1 1991 to Q4 2013. The result of this process is given by (2) and
ur
15
na
(3).15,16
Jo
Actually, the structural design process as described in Hammersland (2017) and summarized in Appendix B was originally undertaken on a three-dimensional structure for the fully simultaneous determination of credit, asset prices and aggregate investment. However, as the investment equation in Appendix B Table B.5 is redundant for the practical implementation of the financial accelerator in KVARTS and the partial model of credit and asset prices in equations (2) and (3) is not only included in the system represented by the equations in Tables B.3- B.5, but also precisely estimated without having to specify the investment equation, it seems safe to assume that aggregate investment is exogenous to the dynamic process determining asset prices and credit, that is, contingent on the long-run structure. This is further substantiated by the χ 2 (5)-test in the Appendix, which fails to reject the lack of a contemporaneous causal link going from investment to either asset prices or credit. However, this does not necessarily mean that the estimated long-run structure of the partial model for asset prices and credit is exogenous for the process driving investment in the long run, which is firmly demonstrated by the fact that there, according to the long-run structure of the system in Table B.3-Table B.5, is a two-way causal link between asset prices and credit on the one hand and investment on the other. This is also why the cointegration analysis referred to in the text was undertaken on a fully simultaneous model for the joint distribution of credit, asset prices and investment in the first place. 16 All estimation and model design in this paper has been made utilizing OxMetrics 7.00 (Doornik, 2007).
11
rk 2nff t 0.04 2 rborsit 0.03jk t 2 0.003RRt 0.24rk 2nff t 4 (2)
0.06rk 2nff 0.77rborsi 0.23 jk t 1
rborsit 2.79rk 2nff t 0.03RRt 0.80mscit 0.27 msci t 1 0.17 mscit 2 0.22 rpoilnok t 0.19 rpoilnok t 1
(3)
ro of
0.46rborsi 0.5rpoilnok 0.5msci 0.04 RR t 1
In equation (2), credit in the long run is determined by Norwegian equity prices and
-p
aggregate investments in fixed capital. Additionally, short-term real interest rates are included in the short term (as well as lags of the dependent variable). Equation (3) shows how
re
Norwegian equity prices can be explained by the real oil price, international equity prices and real interest rates in the short and long term, as well as credit in the short term. See Appendix
lP
B for more detailed estimation results and tests.
Figure 1 illustrates the dynamic projections of the simultaneous structural equation
na
systems (2) and (3) within the estimation period from Q3 2007 to Q4 2013. The fit is
Jo
ur
relatively good throughout this period, even through the financial crisis in autumn 2008.17
The financial sub-model described above is connected to KVARTS via investments, as
determined in the estimated industry-specific equations for capital, where Norwegian equity prices and credit are included as explanatory variables, as explained in the following section. 17
In these dynamic forecasts it has not been necessary to include dummies for the financial crises or other events after the end of the estimation period in the third quarter of 2007. That is, these are the unfettered forecasts of the dynamic structure in (2) and (3) itself.
12
3. Demand for capital 3.1 Modelling of industry investments Industry investments are determined endogenously in KVARTS through 13 industryspecific, estimated equations for real capital. The explanatory variables in these capital equations have traditionally been production and relative factor prices, plus other relevant
ro of
variables such as employment, as shown in
(4) ∆𝑘𝑖,𝑡 = 𝛼𝑖 + ∑5𝑗=1 𝛼𝐾,𝑖,𝑗 ∆𝑘𝑖,𝑡−𝑗 + ∑5𝑗=0 𝛼𝑋,𝑖,𝑗 ∆𝑥𝑖,𝑡−𝑗 + ∑5𝑗=0 𝛼𝑃,𝑖,𝑗 ∆𝑝𝑖,𝑡−𝑗
-p
+ ∑5𝑗=0 𝛼𝑍,𝑖,𝑗 ∆𝑧𝑖,𝑡−𝑗 + 𝛽𝑒𝑐𝑚𝑖,𝑡−1 ,
where ki,t is real capital, xi,t production, pi,t relative factor prices and zi,t represents other
1
lP
solution for real capital is given by
re
relevant variables in industry i, period t. α and β are estimated parameters. The long-term
1
na
(5) 𝑒𝑐𝑚𝑖,𝑡 = 𝑘𝑖,𝑡 + 𝑝𝑖,𝑡 − κ 𝑥𝑖,𝑡 + κ 𝛾𝑖 𝑡𝑟𝑒𝑛𝑑,
ur
where κ is the elasticity of scale and γ captures technological development. These two parameters are estimated in a system, as they are common to the demand of all inputs within
Jo
each industry, see Hungnes (2016). In the present paper, we extend the factor demand relationships (4) by including aggregate credit (c) and equity prices (a) in the short-term dynamics, the latter in addition to the capital cost involved in the long-term relationship through the user cost of capital, so that
13
(6) ∆𝑘𝑖,𝑡 = 𝛼𝑖 + ∑5𝑗=1 𝛼𝑘,𝑖,𝑗 ∆𝑘𝑖,𝑡−𝑗 + ∑5𝑗=0 𝛼𝑥,𝑖,𝑗 ∆𝑥𝑖,𝑡−𝑗 + ∑5𝑗=0 𝛼𝑝,𝑖,𝑗 ∆𝑝𝑖,𝑡−𝑗 + ∑5𝑗=0 𝛼𝑧,𝑖,𝑗 ∆𝑧𝑖,𝑡−𝑗 +∑5𝑗=0 𝛼𝑐,𝑗 ∆𝑐𝑡−𝑗 +∑5𝑗=0 𝛼𝑎,𝑗 ∆𝑎𝑡−𝑗 + 𝛽𝑒𝑐𝑚𝑖,𝑡−1 , where ct is credit to non-financial corporations and at represents the main index on the Oslo Stock Exchange at time t. Note that for credit and equity prices, we use the same aggregate variable in all equations. This is because we do not have data available for these variables at the industry level. This simplification implies assuming that aggregate credit and equity prices are good indicators for all industries, i.e. when the main index on the Oslo stock
ro of
exchange rises and when aggregated credit increases, so does the availability of external financing in general at a disaggregated level.
As credit and equity prices are not included in the long-term relationships in (6), the
-p
level of these variables has in principle no effect in the long run.18 As mentioned, only allowing for direct effects of credit and equity in the short term in the capital equations is in
re
line with the Modigliani-Miller theorem, and implies that in the long term the financial
lP
structure, in other words how companies finance themselves, is not relevant to or dependent on the economic cycle.19 Long-term effects of higher capital prices are, however, partially
na
safeguarded by the user cost of capital.
The investments in industry i in period t, JKi,t, appear as the change in capital stock from the previous period, adjusted for depreciation,
ur
(7) JKit = ∆KAGGi,t - δ*KAGGi,t-1,
Jo
where KAGGi,t is the capital stock and δ is the depreciation rate of capital.
3.2 Empirical results 18
Note that the price of capital is included in the definition of the user cost of capital. However, because there is no link in the model between the user cost of capital and the benchmark index on the Oslo stock exchange, the user cost of capital does not represent a channel for long-term effects between equity prices and the accumulation of capital. 19 Strictly speaking, the Modigliani-Miller theorem applies under relatively strong assumptions of efficient markets and absence of taxes and asymmetric information, and provides little information about adjustments in the short term.
14
The sector-specific equations for capital (6) are estimated by ordinary least squares and reduced using the general to specific methodology (see, e.g., Davidson et al., 1978). It is emphasized that the final estimated relations pass standard statistical tests of serial correlation, heteroscedasticity and normal distribution of the residuals. The estimation of the long-term structure (5) is documented in Hungnes (2016). Conditional on these ecm’s, we reestimate the short short-term dynamics of each industry as shown in (6). Potential path dependency in the reduction process is handled with Autometrics (Doornik, 2009), which is also used to search
ro of
for structural breaks and extreme observations. We have included dummy variables where this has contributed to a theory consistent model specification and/or improved the statistical properties.
-p
Table 1 displays the size of investments in the various industries as a share of total industry investments. In 2014, total industry investments constituted 30.1 percent of total
re
investments in the Norwegian economy and 42.3 percent of investments in mainland Norway.
lP
The table also shows that the capital equations for all industries excluding industry 86 includes credit as an explanatory variable and that equity prices are included in five of the 13
na
equations (see Appendix F for detailed estimation results).20
ur
Jo
The financial sub-model/accelerator can easily be “switched off” by exogenizing credit and Norwegian equity prices. We exploit this in the next section in order to identify the impact of including the financial accelerator in KVARTS.
3.3 Impact of the financial accelerator 20
Whether the data supports including aggregate credit and/or equity prices even in the long term may be an interesting topic for further research.
15
The importance of the financial accelerator is illustrated by three alternative simulations in KVARTS, in addition to a counterfactual experiment where we look at the effect of keeping the level of oil investments and the oil price up over an extended historic period where both quantities were subject to significant downward corrections. First, a permanent increase in the MSCI of 10 percent gives a corresponding increase in Norwegian equity prices during the first two quarters, followed by a relatively rapid decline that gradually decreases in strength, converging towards a long-term increase of almost 5
ro of
percent after 6-7 years. Credit increases markedly during the first years, approaching 4 percent after 6-7 years. Note that fiscal policy and the money market interest rate are exogenously determined, so that the expansionary effects of higher international equity prices
-p
are not offset by tighter economic policies.21 The krone exchange rate, on the other hand, is determined endogenously in the model in all three simulations. Note that this is a partial shift.
re
In reality, one could imagine that increasing international equity prices would go hand
lP
in hand with a global upturn in the business cycle, implying that increased international demand for Norwegian goods and services and increased optimism could push Norwegian investments and equity prices even further. As equity prices and credit are not included
na
directly in the long-term solution of the capital equations, there is only a short-term additional
21
Jo
ur
The rationale for the decision of disregarding endogenous policy responses in the main setup is based on the idea of wanting to purely cultivate the impact of the financial accelerator. However, as is suggested by the simulations in Appendix A, where, in addition to presenting some additional simulations related to the standard case, we look at the case where we have implemented (and switched on) an augmented open economy version of a Taylor rule (Taylor, 1993) in the model, policy rules in general only seem to contribute to moderate the effects of the original shifts studied and to dilute the proper contribution of the financial accelerator. As we primarily want to illustrate the partial effect of changes to some exogenous processes and in this respect to study the role played by the financial accelerator in particular, we have therefore chosen to concentrate on the partial effects of shocks, basing this decision on the absence of policy responses beyond what follows from non-discretionary endogenous behavioral model responses. As far as the need for a fiscal policy response is concerned, in this context it is also important to bear in mind that monetary policy after all is considered as the “first line” defense in coping with economic disturbances. With the relatively modest magnitude of the shocks studied in this paper, an exception granted for the counterfactual experiment in Appendix E, where the lack of a policy response overall probably contributes to overstating the role played by the financial accelerator in promoting shocks, there is therefore reason to assume that the role of fiscal policy would be rather limited. Also, fiscal policy enters the model (KVARTS) in a very complex and detailed way. While this makes it suitable for detailed impact analyses of a number of fiscal instruments, it is also difficult to endogenize in a way that properly takes into account a realistic fiscal policy response.
16
increase in industry investments of 1.1 percent and 2 percent in manufacturing investments, which in both cases disappear within about ten years. GDP Mainland Norway gets an additional increase of 0.2 percent the first year, which decreases slowly towards zero. Unemployment falls rapidly by 0.1 percentage point, but this effect is almost gone after 10 years. The additional effect on traditional exports is very tiny, while imports increase more significantly because investments are relatively import-intensive. Table 2 and Figure 2, together with the graphs in Appendix C, show the effects following from the change in the
ro of
MSCI on a range of macroeconomic variables.22 The graphs in Appendix D also show the effects of a temporary one-year increase in the MSCI index on all the variables listed in Table 2. With the notable exception of credit, they all seem to convey the impression of a very low
-p
degree of persistence in the process of adjustment to temporary shocks.
lP
re
While the international equity price index is a new variable in KVARTS through the financial sub-model, interest rates and fiscal policy are, naturally, important features in earlier
na
versions of the model as well. To identify the importance of the accelerator mechanism for the quantification of the economy’s sensitivity to changes in monetary and fiscal policy, we
ur
perform two times two calculations: For each policy area, we first look at the total effects
Jo
when the financial accelerator is switched off, and then check how the financial accelerator changes this, i.e., the additional effect attributable to the financial accelerator.23
22
Credit and Norwegian equity prices are not deflated in the tables in this report. However, the inflationary effects of the shift in international equity prices, and the additional inflation attributable to the financial accelerator in the interest rates and fiscal policy shifts, are marginal. The real effect is thus very similar to the nominal effect. 23 In addition, Appendix A documents the total effects of the two policy shifts with the financial accelerator switched on, that is compared to the baseline scenario.
17
First, we look at a permanent 1 percentage point reduction of the money market interest rate compared to the baseline scenario. Initially, we keep credit and equity prices exogenous, meaning that the financial accelerator is “switched off”. In this way, we can illustrate KVARTS without a financial accelerator before calculating the additional effect attributable to the financial accelerator when it is switched on. In KVARTS, the money market interest rate affects GDP Mainland Norway through two (close to) equally important channels. First, lower interest rates (relative to international rates) make the krona depreciate.
enterprises.
-p
ro of
That in turn leads to higher import prices and strengthens the competitiveness of Norwegian
re
Thus, exports of goods and services increase. Second, lower interest rates lead to
lP
increasing household consumption and demand for housing, and enterprises in the mainland economy increase investments. Industry investments go up both as a direct result of reduced financing costs and due to increased demand for their products. Employment increases and
na
unemployment falls. Real wages go up slightly after a while. After 7 years, overall industry investments are up by 5.6 percent and GDP Mainland Norway by 1.8 percent, see Table 3.
ur
We then switch on the accelerator mechanism (letting the model determine credit and
Jo
Norwegian stock prices) and find that a permanent 1 percentage point reduction in the threemonth money market rate leads to a short-term additional increase in industry investments of 1.4 percent after two years and in manufacturing investment of 2.4 percent after three years, which can be attributed to the financial accelerator. GDP Mainland Norway is 0.2 percent higher after two years, before the effect gradually diminishes. Thus, while the effects of an
18
increase in international equity prices on the real economy came during the first year, it takes 2-3 years before the maximum effect of the interest rate reduction is achieved. The interest rate is included in the long-term solution of the equity price equation, in the short-term dynamics of the credit equation and in the capital equations via the user cost of capital. Thus, the financial accelerator is engaged, both directly through increased demand for credit and equities and through increased investments due to lower user cost of capital. Furthermore, credit and equity prices are included in the capital equations, which, through
ro of
investments, provide new feedback effects to equity prices and credit, and so on. After about 10 years, the additional effects on the real economy die out. Equity prices are permanently higher because interest rates are included in the long-term solution of (3) and credit is
-p
permanently higher because equity prices are included in the long-term credit relationship in (2). Table 4 and Figure 4, together with the graphs in Appendix C, show the additional effects
re
of the interest rate reduction as a result of incorporating a financial accelerator into KVARTS.
lP
However, as depicted by the graphs pertaining to a temporary one-year interest rate shock in Appendix D, the process of adjustment related to these additional effects are not very persistent for most variables, a notable exception again being the effect on credit.
na
Next, we simulate a permanent increase in government consumption, investment and employment by 1 percent, hereafter called public expenditure for convenience. In this
ur
calculation, the money market rate is also kept unchanged.24 Conversely, if the interest rate
Jo
was determined endogenously, the interest rate relation would “lean against the wind”, responding to increased economic activity by higher interest rates, and thus counteract the expansionary fiscal policy. However, our concern here is how fiscal policy works. First, we increase public expenditure with exogenous credit and equity prices, i.e. with the financial accelerator switched off. Higher public expenditure increases GDP directly by leading to 24
The results related to the effects of an increase in public expenditure when switching on the monetary policy rule in the model are given in Table A.4 and Tab A.5 in Appendix A, respectively. They all seem to convey the impression of a somewhat dampened response to shocks compared to what is the case without a policy rule.
19
higher production in the public administration. That leads to higher demand and thus higher production, even in private industries. Higher production in both public and private sectors increases demand for employment and hence leads to lower unemployment and slightly higher wage growth.
ro of
Household demand increases as a result of higher wages and employment. Consumer prices increase less than nominal wages, resulting in higher real wages. After 10 years, GDP
-p
Mainland Norway is 0.6 percent higher, see Table 5.
re
lP
When we make a new corresponding shift with expansionary fiscal policy, but with the financial accelerator switched on, we find that the accelerator only provides a marginal additional increase in industry investments and GDP, see Table 6 and the graphs in,
na
respectively, Figure 4 and the last part of Appendix C pertaining to the public expenditure shock, for a more detailed documentation. The additional effect is small because the model
ur
has no direct link from public expenditure to financial markets via the equations for equity
Jo
prices and credit, but only an indirect link from increased demand working through the capital equations.25 Looking at the graphs of the additional effect pertaining to a temporary one-year shock to public expenditures – albeit minuscule – reveals a rather substantial degree of persistence in the adjustment process for some of the variables, this time not only for credit.
25
Mazzucato (2015) argues that there can be significant direct effects from public to private investment, including through public/private investment partnerships and so-called bell cow effects.
20
Finally, we look at a counterfactual experiment that is based on a fairly recent experience related to the Norwegian economy and that demonstrates its dependence on oil. In particular, in this context we will look at what would have happened if oil prices and investments, instead of falling precipitously from 2013 onwards and staying low for an extended period of time thereafter, had stayed at their level before the shock took place at the
ro of
end of 2013, beginning of 2014, over the simulation period. As shown in the memo of Table 7, such a shock would amount to a gradual increase in oil prices and oil investments relative to the baseline scenario of close to, respectively, 150 and 20 percent over a four- to five-year
-p
period, before settling down at approximately 80 and 16 percent at the end of the simulation period in 2020. As borne out by Table 7, which gives us the additional effects of the
re
counterfactual shift attributable to the inclusion of a financial accelerator, the relative
lP
importance of the financial accelerator in wake of the shocks can be quite substantial. In Table 7 this additional effect is simulated to amount to an additional boost to GDP of close to 1 percent in 2017 before settling down to about 0,6 percent at the end of the simulation
na
period. This process of enhancement is eventually propagated through a substantial upturn on the Oslo stock exchange and a subsequent strong increase in credit, both rendering possible a
Jo
ur
cyclical upturn in investment and consumption.
5. Conclusion We have estimated and implemented a financial sub-model in a macro-econometric model of the Norwegian economy (KVARTS), which takes into account the pro-cyclical
21
interaction between the real economy and financial markets via industry investments. Our implementation is more theory-consistent than previous studies, as the financial variables affect investments directly and we have taken into account that the effects of changing credit and equity prices on investments can be industry-specific. In the financial sub-model, aggregated credit and Norwegian equity prices are determined simultaneously in a twodimensional structural system characterized by full contemporaneous causal interaction, where industry investments are included as an explanatory variable. Furthermore, aggregate
ro of
credit and Norwegian equity prices are included as explanatory variables in the equations for total capital formation in each industry. The industry-specific real capital is aggregated to total capital formation in the mainland industries, which is included in the dynamics of the
-p
financial sub-model. In this way, equity prices and credit affect investments with feedback effects to equity prices and credit and so on.
re
The impact of the financial accelerator is illustrated by three shifts in exogenous
lP
variables in addition to a counterfactual experiment. A permanent reduction in three-month money market interest rates of 1 percentage point provides a short-term additional increase in industry investments by 1.4 percent, attributable to the financial accelerator. A permanent
na
increase in public expenditure of 1 percent provides just a marginal additional increase in industry investments as a result of the financial accelerator, as there is no direct connection
ur
from public expenditure to the financial accelerator. We also shifted the global equity price
Jo
index MSCI, which is included in both the short-term dynamics and the long-term structure in the equation for Norwegian equity prices. A permanent increase in the MSCI of 10 percent provides a long-term increase in Norwegian equity prices of about 5 percent and a short-term increase of about 1 percent only in industrial investments. As far as the counterfactual experiment is concerned, the additional effect that is attributed to the inclusion of the financial variable is simulated to give an additional boost to GDP of close to 1 percent. This is
22
taken as clear evidence of the intensive role played by a financial accelerator in the propagation of shocks. Including the financial sub-model in KVARTS reinforces and extends economic cycles in projections and forecasts for the Norwegian economy. Moreover, monetary policy gets a more important role, as the effect of a change in interest rates is significantly enhanced. The effects of fiscal policy, on the other hand, are affected to a relatively small extent as a
Jo
ur
na
lP
re
-p
ro of
result of this model extension.
23
References Adrian, T., P. Colla and H. S. Shin (2013): “Which Financial Frictions? Parsing the Evidence from the Financial Crisis of 2007-9”: NBER Macroeconomics Annual 2012, Volume 27 (2013), Acemoglu, D., J. Parker and M. Woodford (red.), p. 159 – 214. Elsevier.
Anundsen, A. K. and E. S. Jansen (2013): “Self-reinforcing effects between housing prices and credit”. Journal of Housing Economics, 22(3), p. 192–212.
ro of
Benedictow, A. and P. Boug (2012): “Trade liberalisation and exchange rate pass-through: the case of textiles and wearing apparels”. Empirical Economics 45, p. 757–788.
Bernanke, B. and M. Gertler (1989): “Agency costs, net worth and business fluctuation”.
-p
American Economic Review 79(1), p. 14–31.
re
Bernanke, B., M. Gertler and S. Gilchrist (1999): “The financial accelerator in a quantitative business cycle framework”. Taylor, J. and M. Woodford (red), Handbook of
lP
Macroeconomics, Vol. 1, Elsevier Science B.V, p. 1341–1393.
Boug, P., Cappelen Å. and A. R. Swensen (2006): “Expectations and regime robustness in
na
price formation: evidence from vector autoregressive models and recursive methods”.
ur
Empirical Economics 31, p. 821–845.
Boug, P. and A. Fagereng (2010): “Exchange rate volatility and export performance: a
Jo
cointegrated VAR approach”. Applied Economics 42, p. 851–864
Boug, P., Å. Cappelen and T. Eika (2013A): “The importance of the distribution sector for exchange rate pass-through in a small open economy: a large scale macroeconometric modeling approach”. Discussion Papers 731, Statistics Norway, Appendix A.
24
Boug, P., Å. Cappelen and T. Eika (2013B): “Exchange rate pass-through in a small open economy: the importance of the distribution sector”. Open Economic Review 24, p. 853 - 879. Bowitz, E., Å. Cappelen (2001): “Modeling income policies: some Norwegian experiences 1973-1993”. Economic Modelling 18, p. 349–379. Brunnermeier, M. K. (2001): “Asset Pricing Under Asymmetric Information: Bubbles, Crashes, Technical Analysis, and Herding”. Oxford University Press.
ro of
Brunnermeier, M. K. (2008): ”Bubbles”, in New Palgrave Dictionary of Economics, ed. by S. N. Durlauf, and L. E. Blume. Palgrave Macmillan.
Brunnermeier, M. K., T. M. Eisenbach and Y. Sannikov (2012): “Macroeconomics with
-p
financial frictions: A survey”. NBER Working Paper No. 18102.
re
Bårdsen, G. and R. Nymoen (2009): “Macroeconometric modeling for policy”. Mills, T. C. and K. Patterson (red.), Palgrave Handbook of Econometrics Vol. 2, 17, p. 851–916.
lP
Palgrave-Macmillan.
Christensen, I. and A. Dib (2008): “The financial accelerator in an estimated New Keynesian
na
model”. Review of Economic Dynamics, Vol 1, No 1, p. 155–178.
ur
Davidson, J. E. H., D. F. Hendry, F. Srba, and S. Yeo (1978): “Econometric Modelling of the Aggregate Time-series Relationship Between Consumers’ Expenditure and Income in the
Jo
United Kingdom”. Economic Journal, 88, 352, p. 661—692.
Diamond, P. and H. Vartiainen (2012): “Behavioral Economics and its Applications”. Princeton University Press.
Doornik, J. A. (2007): “Object-Oriented Matrix Programming Using Ox”. 3rd ed. London: Timberlake Consultants Press and Oxford. 25
Doornik, J. A. (2009): “Autometrics”. Castle, J. L. and N. Shepard (2009): “The methodology and practice of econometrics”, Oxford university press.
Fama, E. (1965): “The behavior of stock market prices”. The Journal of Business, Vol. 38, No. 1, p. 34–105. The University of Chicago Press.
ro of
Friedman, M. (1957): “A theory of the consumption function”. Princeton University Press.
-p
Hall, R. E. (2011): “The High Sensitivity of Economic Activity to Financial Frictions”. The
re
Economic Journal 121(552), p. 351–378.
Hammersland, R. (2017): “The Financial Accelerator and the Real Economy: Evidence Using
lP
a data based procedure of Simultaneous Structural Model Design”. Proceedings ITISE 2017
na
(2), p. 1007- 1034.
Hammersland, R. and C. B. Træe (2014): “The financial accelerator and the real economy: A
ur
small macroeconometric model for Norway with financial frictions”. Economic Modeling,
Jo
vol. 36(C), p. 517–537.
Hartley, J., K. Hoover and K. D. Salyer (1998): “Real business cycles: a reader”. Routledge. Hirano, T. and N. Yanagawa (2016):”Asset Bubbles, Endogenous Growth, and Financial Frictions”. Review of Economic Studies 84 (1), p. 406-443.
26
Hubbard, R.G. (1998): “Capital market imperfections and investment”. Journal of Economic Literature 36, p. 193–225. Hungnes, H. (2016): “Using common factors to identify substitution possibilities in a factor demand system with technological changes”. Discussion papers No. 849, Statistics Norway.
Jansen E. S. (2013): “Wealth effects on consumption in financial crisis: the case of Norway”.
ro of
Empirical Economics 45, p. 873–904.
Jermann, U. and V. Quadrini (2012): “Macroeconomic Effects of Financial Shocks”.
-p
American Economic Review, 102(1), p. 238-271.
Johansen, L. (1995): “Likelihood-Based Inference in Cointegrated Vector Autoregressive
re
Models”. New York. Oxford University Press.
lP
Kiyotaki, N. and J. Moore (1997): “Credit cycles”. Journal of Political Economy,
na
105(2), p. 211–248.
Kydland, F. E. and E. C. Prescott (1982): “Time to build and aggregate fluctuations,”
ur
Econometrica 50(6), p. 1345–1369.
Jo
Mazzucato, M. (2015): Building the Entrepreneurial State: A New Framework for Envisioning and Evaluating a Mission-oriented Public Sector”. Wp 824, Levy Economics Institute.
Modigliani, F. and M. Miller (1958): “The Cost of Capital, Corporation Finance and the Theory of Investment”. American Economic Review 48 (3), p. 261–297.
27
Silvestrini, A. and A. Zaghinib (2015): “Financial shocks and the real economy in a nonlinear world: From theory to estimation”. Journal of Policy Modelling 37(6), p. 915–929.
Smets, F. and R. Wouters (2007): “Shocks and frictions in US business cycles: A Bayesian DSGE approach”. American Economic Review, 97 (3), p. 586–606.
ro of
Stiglitz, J. E. (1982): “The Inefficiency of the Stock Market Equilibrium”. Review of Economic Studies, 49(2), p. 241–261.
-p
Stiglitz, J. E., and A. Weiss (1981): “Credit Rationing in Markets with Imperfect
re
Information”. American Economic Review, 71(3), p. 393–410.
lP
Stiglitz, J. E. (2010): “Freefall”. W. W. Norton & Company, New York.
na
Taylor, John B. (1993). "Discretion versus Policy Rules in Practice" (PDF). CarnegieRochester Conference Series on Public Policy. 39: 195–214.
Tobin, J. (1969): “A general equilibrium approach to monetary theory”. Journal of Money
Jo
ur
Credit and Banking 1, p. 15–29.
28
Jo
ur
na
lP
re
-p
ro of
Figure 1. Credit (rk2nff) and Norwegian equities (rborsi), with corresponding, model-based projections
29
Figure 2. Macroeconomic effects of a permanent increase in the MSCI by 10 percent. Deviations from the baseline scenario in percent (corresponding to Table 2) 10
1.5 1.3 1.0 0.8 0.5 0.3 0.0
8 6 4 2 0 2
3
4
5
6
7
8
9 10
1
2
3
4
5
6
7
8
9 10
Oslo benchmark share index
GDP Mainland Norway
Credit to non-financial industries
Industry investments
ro of
1
Figure 3. Macroeconomic effects of a permanent reduction in the Norwegian money market rate of 1 percentage point of incorporating a financial accelerator. Deviations from the model without financial accelerator in percent (corresponding to Table 4) 1.5 1.3 1.0 0.8 0.5 0.3 0.0
10
6 4
re
2
-p
8
0
1
2
3
4
5
6
7
8
9 10
2
3
4
5
6
7
8
9 10
Oslo benchmark share index
lP
GDP Mainland Norway
1
na
Industry investments
Credit to non-financial industries
Figure 4. Macroeconomic effects of a permanent increase in public expenditure by 1 percent of incorporating a financial accelerator. Deviations from the model without financial accelerator in percent (corresponding to Table 6) 10
ur
1.50 1.25 1.00 0.75 0.50 0.25 0.00
8 6
Jo
4 2 0
1 2 3 4 5 6 7 8 9 10
1
2
3
4
5
6
7
8
9 10
GDP Mainland Norway
Oslo benchmark share index
Industry investments
Credit to non-financial industries
30
Table 1. Investment in each industry as a share of total industry investments (value). Inclusion of credit and/or equity prices as explanatory variables in the short-term dynamics of each equation is indicated by X.
Share
Credit
Industry 10 – Agriculture etc.
3.6
X
Industry 14 – Fishing and hunting
1.2
X
Industry 15 – Consumer goods
3.6
X
Industry 25 – Intermediate goods etc.
3.2
X
Industry 30 – Energy-intensive goods
2.5
X
Equity prices
X
X
ro of
Total industry investments in 2014
Industry 45 – Engineering products Industry 55 – Construction Industry 63 - Banking and insurance Industry 71 – Electricity
-p
Industry 74 – Domestic transport Industry 81 – Merchandising
Industry 86 - Leasing commercial buildings Total industry investments
lP
Industry investments as a share of total investments
re
Industry 85 - Other private services
Jo
ur
na
Industry investments as a share of total investments in Mainland Norway
31
6.1
X
7.1
X
3.6
X
8.8
X
9.3
X
6.5
X
27.9
X
16.5 100 30.1 42.3
X
X X
Table 2. Macroeconomic effects of a permanent increase in the MSCI by 10 percent. Deviations from the baseline scenario in percent
2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.2
0.2
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
Household consumption
0.3
0.3
0.3
0.3
0.4
0.3
0.3
0.3
0.2
0.2
-0.1
-0.1
-0.1
-0.1
-0.1
-0.1
-0.1
-0.1
0.0
0.0
0.5
0.5
0.5
0.4
0.4
0.3
0.3
0.3
0.2
0.2
Industries
1.1
1.1
0.9
0.7
0.5
0.4
0.3
0.2
0.2
0.1
Manufacturing
1.3
2.0
1.6
1.3
0.8
0.6
0.4
0.3
0.2
0.2
Exports traditional goods
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Imports
0.3
0.3
0.3
0.3
0.3
0.3
0.2
0.2
0.2
0.1
Wage
0.0
0.0
0.0
0.1
0.1
0.1
0.1
0.1
0.1
0.1
CPI
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.7
1.9
2.8
3.3
9.5
7.3
6.2
5.6
10.0
10.0
10.0
1
Oslo Børs benchmark index Memo
10.0
Jo
ur
na
lP
MSCI 1 Nominal
-p
Unemployment rate (level) Investment in Mainland Norway
ro of
1
3.6
3.7
3.8
3.8
3.8
3.8
5.3
5.1
4.9
4.9
4.9
4.8
10.0
10.0
10.0
10.0
10.0
re
Year
32
10.0
Table 3. Macroeconomic effects without financial accelerator of a permanent reduction in the Norwegian money market rates by 1 percentage point. Deviations from the baseline scenario in percent 2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.3
0.8
1.1
1.3
1.5
1.7
1.8
1.8
1.8
1.7
Household consumption
0.1
0.7
1.2
1.6
2.0
2.2
2.3
2.3
2.2
2.1
-0.7
-0.8
-1.0
-1.2
-1.3
-1.6
-1.7
-1.7
-1.6
-1.4
0.5
1.6
2.3
3.3
4.2
5.1
5.7
5.7
5.5
5.1
Industries
1.0
3.4
3.8
4.3
4.7
5.3
5.6
5.5
5.2
4.8
Manufacturing
0.6
2.7
3.7
3.9
3.4
3.6
3.8
3.8
3.5
3.3
Exports traditional goods
1.6
1.7
1.9
1.8
1.6
1.4
1.3
1.2
1.1
1.1
Imports
0.1
0.7
1.1
1.4
1.7
2.0
2.2
2.2
2.1
1.9
Wage
0.3
0.6
0.8
1.1
1.4
1.6
1.8
1.9
2.1
2.2
CPI
0.4
0.7
0.8
0.9
1.0
1.0
1.1
1.0
1.0
0.9
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
4.4
4.6
4.2
3.9
3.7
3.6
3.5
3.3
3.2
3.0
0.3
0.7
0.9
1.1
1.2
1.4
1.4
1.5
1.6
1.7
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
-1.0
-1.0
-1.0
1
Oslo Børs benchmark index
-1.0
Jo
ur
na
lP
Memo Money market interest rate (percentage points) 1 Nominal
-p
Unemployment rate (level) Investment in Mainland Norway
33
ro of
1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
-1.0
-1.0
-1.0
-1.0
-1.0
re
Year
-1.0
Table 4. Macroeconomic effects of a permanent reduction in Norwegian money market rates by 1 percentage point of incorporating a financial accelerator. Deviations from the effects without a financial accelerator in percent (ref. Table 3) 2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.1
0.2
0.2
0.2
0.2
0.1
0.1
0.1
0.1
0.1
Household consumption
0.1
0.3
0.3
0.3
0.4
0.3
0.3
0.2
0.2
0.2
Unemployment rate (level) Investment in Mainland Norway
0.0
-0.1
-0.1
-0.1
-0.1
-0.1
-0.1
-0.1
0.0
0.0
0.3
0.6
0.7
0.6
0.5
0.5
0.4
0.3
0.2
0.2
Industries
0.6
1.4
1.3
1.2
1.0
0.8
0.6
0.4
0.3
0.1
Manufacturing
0.9
2.2
2.4
2.2
1.7
1.3
0.9
0.6
0.4
0.3
Exports traditional goods
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Imports
0.1
0.3
0.3
0.3
0.3
0.3
0.3
0.2
0.2
0.1
Wage
0.0
0.0
0.0
0.1
0.1
0.1
0.1
0.1
0.1
0.1
CPI
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.0
0.6
2.0
3.3
4.4
Oslo Børs benchmark index1 1 Nominal
4.4
7.8
8.3
7.6
-p 5.1
5.6
5.8
5.9
6.0
5.9
7.1
6.6
6.3
5.9
5.7
5.5
re
na ur Jo 34
ro of
1
lP
Year
Table 5. Macroeconomic effects without financial accelerator of a permanent increase in public expenditure by 1 percent. Deviations from the baseline scenario in percent 2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.3
0.4
0.4
0.4
0.5
0.5
0.6
0.6
0.6
0.6
Household consumption
0.1
0.3
0.4
0.5
0.6
0.6
0.7
0.7
0.7
0.7
Unemployment rate (level) Investment in Mainland Norway
-0.6
-0.4
-0.4
-0.4
-0.4
-0.5
-0.6
-0.6
-0.6
-0.6
0.3
0.3
0.5
0.6
0.8
1.0
1.1
1.2
1.2
1.2
Industries
0.1
0.3
0.3
0.4
0.5
0.6
0.7
0.8
0.7
0.7
Manufacturing
0.0
0.1
0.1
0.1
0.1
0.1
0.2
0.2
0.2
0.2
Exports traditional goods
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Imports
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.8
0.8
0.8
Wage
0.2
0.2
0.3
0.4
0.5
0.5
0.6
0.7
0.7
0.8
CPI
0.0
0.0
0.0
0.1
0.1
0.1
0.2
0.2
0.2
0.3
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.2
0.2
0.2
0.3
0.3
0.4
0.5
0.5
0.5
0.6
0.6
0.6
0.6
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.0
1.0
1.0
Oslo Børs benchmark index Memo
1.0
Jo
ur
na
lP
Public expenditure 1 Nominal
-p
1
35
ro of
1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.0
1.0
1.0
1.0
1.0
re
Year
1.0
Table 6. Macroeconomic effects of a permanent increase in public expenditure by 1 percent of incorporating a financial accelerator. Deviations from the effects without financial accelerator in percent (ref. Table 5) 2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.00
0.00
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
Household consumption
0.00
0.01
0.01
0.01
0.02
0.02
0.02
0.02
0.02
0.02
Unemployment rate (level) Investment in Mainland Norway
0.00
0.00
0.00
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
0.01
0.02
0.03
0.03
0.04
0.04
0.05
0.05
0.04
0.04
Industries
0.01
0.05
0.06
0.07
0.08
0.09
0.09
0.09
0.09
0.08
Manufacturing
0.03
0.09
0.12
0.14
0.14
0.15
0.15
0.15
0.15
0.14
Exports traditional goods
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Imports
0.00
0.01
0.01
0.01
0.02
0.02
0.02
0.02
0.02
0.02
Wage
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.01
0.01
CPI
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.02
0.08
0.15
0.22
0.05
0.15
0.22
1
0.27
Jo
ur
na
lP
Oslo Børs benchmark index 1 Nominal
-p
ro of
1
0.01
0.01
0.01
0.01
0.01
0.01
0.30
0.37
0.45
0.52
0.60
0.66
0.30
0.34
0.39
0.42
0.46
0.50
re
Year
36
Table 7. Macroeconomic effects of a countercyclical shock to oil prices and oil investments. Deviations from the effects without a financial accelerator in percent 2013
2014
2015
2016
2017
2018
2019
2020
GDP Mainland Norway
0.01
0.05
0.38
1.03
1.10
0.91
0.74
0.61
Household consumption
0.03
0.1
0.85
2.26
2.36
1.97
1.71
1.48
-0.01
-0.03
-0.2
-0.54
-0.7
-0.71
-0.63
-0.45
0.04
0.15
1.04
2.9
3.45
3.08
2.47
1.9
Industries
0.09
0.35
2.48
7.07
8.4
7.13
5.23
3.37
Manufacturing
0.04
0.41
2.57
9.46
13.5
12.26
9.05
5.73
Exports traditional goods
0.00
0.00
-0.01
-0.05
-0.13
-0.20
-0.23
-0.21
Imports
0.23
0.08
0.69
1.91
2.13
1.86
1.6
1.31
Wage
0.00
0.01
0.03
0.11
0.23
0.36
0.48
0.58
-0.00
-0.00
-0.02
-0.07
-0.07
-0.03
0.03
0.08
0.00
-0.00
-0.01
-0.06
-0.08
-0.04
0.01
0.07
0.01
0.03
0.17
0.32
0.49
0.54
0.57
0.55
0.03
0.28
1.98
0.95
4.2
28.08
3.499
13.10
0.41
0.48
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1 1
Oslo Børs benchmark index Memo: Oil price
31.43
109.91
155.48
122.87
106.51
92.39
80.08
5.67
15.87
21.66
19.94
17.09
16.06
Jo
ur
na
lP
Oil investments 1 Nominal
-p
CPI
8.5
17.78
26.04
33.89
65.66
70.83
61.5
50.66
41.42
re
Unemployment rate (level) Investment in Mainland Norway
ro of
Year
37
Appendix A. Macroeconomic effects of economic policy Without a monetary policy rule Table A.1. Macroeconomic effects including a financial accelerator of a permanent reduction in the Norwegian money market rates by 1 percentage point. Deviations from the baseline scenario in percent. År
2
3
4
5
6
7
8
9
10
0.9
1.2
1.5
1.7
1.9
2.0
1.9
1.9
1.7
0.2
1.0
1.5
1.9
2.3
2.5
2.6
2.6
2.4
2.2
-0.7
-0.9
-1.1
-1.3
-1.4
-1.7
-1.8
-1.8
-1.6
-1.4
0.8
2.3
3.0
3.9
4.7
5.6
6.1
6.1
5.8
5.2
Industries
1.6
4.8
5.2
5.6
5.8
6.2
6.3
5.9
5.4
5.0
Manufacturing
1.6
5.0
6.2
6.2
5.2
5.0
4.8
4.4
3.9
3.6
Exports traditional goods
1.6
1.7
1.9
1.8
1.5
1.4
1.2
1.2
1.1
1.1
Imports
0.3
1.0
1.4
1.7
2.0
2.3
2.4
2.4
2.2
2.0
Wage
0.3
0.6
0.9
1.2
1.5
1.7
CPI
0.4
0.7
0.8
0.9
1.0
1.1
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
4.4
4.6
4.2
3.9
3.7
3.6
0.3
0.8
1.0
1.2
1.4
1.5
0.6
2.0
3.5
4.6
5.4
5.9
4.6
8.5
9.0
8.2
-1.0
-1.0
-1.0
-1.0
Unemployment rate (level) Investment in Mainland Norway
1
Oslo Børs benchmark index
1.9
2.0
2.2
2.4
1.1
1.0
1.0
0.9
3.5
3.3
3.2
3.1
1.5
1.6
1.7
1.8
6.2
6.3
6.3
6.3
7.6
7.1
6.7
6.3
6.0
5.8
-1.0
-1.0
-1.0
-1.0
-1.0
-1.0
re
Memo Money market interest rate (percentage points) 1 Nominal.
ro of
Household consumption
1 0.4
-p
GDP Mainland Norway
År GDP Mainland Norway Household consumption
Industries Manufacturing Imports Wage
4
5
6
7
8
9
10
0.5
0.5
0.5
0.6
0.6
0.6
0.6
0.3
0.4
0.5
0.6
0.7
0.7
0.7
0.7
0.7
-0.4
-0.4
-0.4
-0.4
-0.6
-0.6
-0.6
-0.6
-0.6
0.4
0.5
0.7
0.9
1.0
1.2
1.2
1.3
1.2
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.8
0.8
0.8
0.0
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.8
0.8
0.8
0.2
0.2
0.3
0.4
0.5
0.5
0.6
0.7
0.7
0.8
0.0
0.0
0.0
0.1
0.1
0.1
0.2
0.2
0.2
0.3
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.2
0.2
0.2
0.3
0.3
0.4
0.5
0.5
0.5
0.6
0.6
0.6
0.6
0.0
0.1
0.1
0.2
0.3
0.4
0.4
0.5
0.6
0.7
Oslo Børs benchmark index1
0.0
0.2
0.2
0.3
0.3
0.3
0.4
0.4
0.5
0.5
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
Jo
CPI
3
0.4
0.3
ur
Exports traditional goods
0.1 -0.6
2
0.4
na
Unemployment rate (level) Investment in Mainland Norway
1 0.3
lP
Table A.2. Macroeconomic effects including a financial accelerator of a permanent increase in government spending by 1 percent. Deviations from the baseline scenario in percent.
Memo Public expenditure 1 Nominal.
38
With a monetary policy rule Table A.3. Macroeconomic effects of a permanent increase in the MSCI by 10 percent. Deviations from the baseline scenario in percent.
Year
1
2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.16
0.15
0.15
0.13
0.13
0.12
0.1
0.08
0.06
0.05
Household consumption
0.33
0.31
0.29
0.33
0.35
0.31
0.26
0.21
0.17
0.14
-0.09
-0.09
-0.1
-0.10
-0.09
-0.08
-0.06
-0.04
0.02
0.0
0.51
0.5
0.45
0.39
0.33
0.29
0.24
0.18
0.13
0.06
Industries
1.08
1.09
0.89
0.69
0.49
0.32
0.2
0.12
0.07
0.05
Manufacturing
1.29
2.02
1.63
1.22
0.76
0.49
0.28
0.18
0.16
0.16
Unemployment rate (level) Investment in Mainland Norway
0.0
0.0
-0.01
-0.02
-0.02
-0.02
-0.01
-0.01
0.0
0.28
0.28
0.28
0.28
0.28
0.24
0.20
0.16
0.12
0.08
Wage
0.02
0.03
0.04
0.06
0.07
0.08
0.08
0.08
0.07
0.07
CPI
-0.00
-0.01
-0.01
-0.01
-0.0
-0.0
0.00
0.01
0.02
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1 Oslo Børs benchmark index1
-0.00
-0.00
-0.03
-0.05
-0.06
-0.06
-0.06
-0.05
0.01 -0.026
-0.00
0.09
0.13
0.14
0.13
0.13
0.09
0.06
0.05
0.04
0.04
0.73
1.89
2.8
3.32
3.58
3.7
3.74
3.74
3.72
3.71
9.52
7.27
6.20
5.57
5.15
4.91
4.79
4.74
4.74
4.76
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
Jo
ur
na
lP
MSCI 1 Nominal.
re
Memo
39
ro of
0.0
Imports
-p
Exports traditional goods
Table A.4. Macroeconomic effects without financial accelerator of a permanent increase in public expenditure by 1 percent. Deviations from the baseline scenario in percent. Year GDP Mainland Norway Household consumption Unemployment rate (level) Investment in Mainland Norway Industries
1
2
3
4
5
6
7
8
9
10
0.3
0.3
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.1
0.2
0.3
0.4
0.5
0.5
0.5
0.5
0.5
0.5
-0.5
-0.3
-0.3
-0.4
-0.3
-0.4
-0.5
-0.5
-0.5
-0.5
0.3
0.3
0.4
0.5
0.6
0.7
0.8
0.8
0.8
0.7
0.1
0.1
0.1
0.2
0.3
0.3
0.3
0.3
0.2
0.2
0.0
0.0
0.1
0.1
0.1
0.1
0.2
0.2
0.2
0.2
-0.1
-0.1
-0.1
-0.1
-0.1
-0.2
-0.2
-0.2
-0.2
-0.2
Imports
0.2
0.3
0.4
0.5
0.5
0.6
0.6
0.6
0.6
0.6
Wage
0.1
0.2
0.3
0.3
0.4
0.5
0.5
0.5
0.6
0.6
CPI
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.2
-0.1
-0.2
-0.2
-0.2
-0.2
-0.2
-0.3
-0.3
-0.2
-0.2
0.3
0.3
0.4
0.4
0.5
0.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.0
1.0
1.0
1.0
1.0
1.0
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1 Oslo Børs benchmark index1
Jo
ur
na
lP
re
Public expenditure 1 Nominal.
40
0.5
0.5
0.5
0.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.0
1.0
1.0
1.0
-p
Memo
ro of
Manufacturing Exports traditional goods
Table A.5. Macroeconomic effects of a permanent increase in public expenditure by 1 percent of incorporating a financial accelerator. Deviations from the effects without financial accelerator in percent (ref. Table 5 for a comparison).
2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Household consumption
0.00
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
Unemployment rate (level) Investment in Mainland Norway
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
-0.01
-0.00
0.00
-0.00
-0.00
-0.00
-0.01
-0.00
-0.00
Industries
0.00
-0.02
-0.00
0.00
-0.00
-0.00
-0.00
-0.01
-0.00
0.01
Manufacturing
0.02
0.01
0.01
0.02
0.01
0.01
0.01
0.00
0.01
0.01
Exports traditional goods
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Imports
0.00
-0.01
-0.00
-0.00
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
Wage
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.01
0.01
CPI
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.02
Oslo Børs benchmark index1
-0.07
-0.21
-0.17
-0.18
1.0
1.0
1.0
1.0
Jo
ur
na
lP
Public expenditure 1 Nominal.
-p
Memo
41
ro of
1
0.02
0.03
0.03
0.03
0.04
0.05
-0.24
-0.27
-0.31
-0.34
-0.33
-0.31
1.0
1.0
1.0
1.0
1.0
re
Year
1.0
Appendix B. Financial sub-model Variables. Source: Statistics Norway, except MSCI: Macrobond. BORSI = Benchmark stock index Oslo (OSEBX) K2NFF = Credit to non-financial corporations in Mainland-Norway JK = I = Gross fixed capital formation REALR = Real interest rate MSCI = Morgan Stanley Capital International: Broad global equity benchmark index, see www.msci.com POILNOK = Oil price, Brent blend PYMN = Deflator GDP Mainland-Norway PJK = Price index for investments / user costs RC= Real repurchasing costs Algebra.
ro of
A = BORSI/PYMN; a = log(A); Da = diff(a,1); D4a = diff(a,4); DDa = diff(Da,1);
-p
RC=PJK/PYMN; rc=log(RC);
re
pymn = log(PYMN); Dpymn= diff(pymn,1);
lP
c = log(K2NFF /PYMN); Dc = diff(c,1);
na
RPOILNOK = POILNOK/PYMN; rpoilnok = log (RPOILNOK); Drpoilnok = diff(rpoilnok,1); DREALR = diff(REALR,1)
ur
i = log(I); Di = diff(i,1);
Jo
msci=log(MSCI); Dmsci =diff(msci,1);
DUMYYQ = Dummy equal to 1 in year YY and quarter Q, zero otherwise. IDUMYYQ = Impuls dummy equal to 1 -1 in respectively quarter Q and Q+1, year YY.
42
Table B.1. Equation for ∆c. CFIML: 1990(4)-2013(4) Coefficient
Std.
t-value
t-prob
CIa_1
-0,05595
0,006704
-8,34
0
Constant
0,635864
0,0755
8,42
0
∆i_2
0,023965
0,01374
1,74
0,0873
∆a
0,036147
0,01056
3,42
0,0012
∆a_1
-0,03615
0,01056
-3,42
0,0012
∆REALR
-0,0031
0,002531
-1,23
0,2263
∆c_4
0,174756
0,06565
2,66
0,0104
DUM933
-0,03055
0,01424
-2,14
0,0368
sigma = 0,0151042
Table B.2. Equation for ∆a. CFIML: 1990(4)-2013(4) Std.
t-value
t-prob
CIb_1
-0,4597
0,06037
-7,61
0
∆c
2,79348
0,5184
5,39
0
Constant
-3,13626
0,4109
-7,63
0
∆REALR
-0,0272
0,01174
-2,32
0,0247
∆msci
0,792869
0,08321
9,53
0
∆msci_1
0,268681
0,07605
3,53
0,0009
∆msci_2
0,166451
0,07735
2,15
0,0362
∆rpoilnok
0,224819
0,05797
3,88
0,0003
∆rpoilnok_1
-0,19894
0,05225
-3,81
0,0004
DUM08Q4
-0,19296
0,06962
-2,77
0,0078
IDUM08Q1
-0,08674
0,0428
-2,03
0,048
re
lP
na
sigma = 0,0694159
-p
Coefficient
ro of
CIa = c-0,770716*a-0,229284*i
CIb = a+0,0351764*REALR-0,493792*msci-0,506208*rpoilnok
ur
BFGS using analytical derivatives (eps1=0.0001; eps2=0.005): Strong convergence log-likelihood 389.109982 no. of observations 93
Jo
Vector SEM-AR 1-5 test: Vector Normality test: Vector Hetero test:
-T/2log|Omega| no. of parameters
F(20,146) = Chi^2(4) = F(216,54) =
43
653.032549 18
1.1069 [0.3486] 3.5719 [0.4670] 0.98715 [0.5411]
Table B.3. Equation for ∆c. CFIML: 1990(4)-2013(4) Coefficient
Std.
t-value
t-prob
CIa_1
-0,05595
0,006704
-8,34
0
Constant
0,635864
0,0755
8,42
0
∆i_2
0,023965
0,01374
1,74
0,0873
∆a
0,036147
0,01056
3,42
0,0012
∆a_1
-0,03615
0,01056
-3,42
0,0012
-0,0031
0,002531
-1,23
0,2263
∆c_4
0,174756
0,06565
2,66
0,0104
DUM933
-0,03055
0,01424
-2,14
0,0368
∆REALR
sigma = 0,0151042
Table B.4. Equation for ∆a. CFIML: 1990(4)-2013(4) Coefficient
Std.
t-value
t-prob
CIb_1
-0,4597
0,06037
-7,61
0
∆c
2,79348
0,5184
5,39
0
-3,13626
0,4109
-7,63
0
Constant ∆REALR
0,01174
-2,32
0,0247
0,792869
0,08321
9,53
0
∆msci_1
0,268681
0,07605
3,53
0,0009
∆msci_2
0,166451
0,07735
2,15
0,0362
∆rpoilnok
0,224819
0,05797
3,88
0,0003
∆rpoilnok_1
-0,19894
0,05225
-3,81
0,0004
DUM08Q4
-0,19296
0,06962
-2,77
0,0078
IDUM08Q1
-0,08674
0,0428
-2,03
0,048
lP
re
-p
-0,0272
∆msci
ro of
CIa = c-0,770716*a-0,229284*i
sigma = 0,0694159
na
CIb = a+0,0351764*REALR-0,493792*msci-0,506208*rpoilnok
Table B.5. Equation for ∆i. CFIML: 1990(4)-2013(4) Coefficient
Std.
t-value
t-prob
0,06821
-4,15
0
3,41037
0,8191
4,16
0
-0,252202
0,09309
-2,71
0,0092
0.186126
0.07937
2.35
0.0229
0,082196
0,05108
1,61
0,1138
--0.050786
0.02846
-1.78
0.0803
CS
-0,122141
0,02436
-5,0
0,0000
CS_1
-0,081953
0,01929
-4,25
0,0001
CS_2
-0,059541
0,01755
-3,29
0,0013
Constant ∆i_1
Jo
∆i_4
-0,283233
ur
CIc_1
∆a
∆4a_1
sigma = 0,0517617 CIc = i-0,60573*(a-rc)
BFGS using analytical derivatives (eps1=0.0001; eps2=0.005): 44
Strong convergence log-likelihood: 540.610624 no. of observations 93 Vector SEM-AR 1-5 test: Vector Normality test: Vector Hetero test:
-T/2log|Omega| no. of parameters
F(45,199) = Chi^2(6) = F(420,110)=
936.494475 27
1.0070 [0.4685] 4.0047 [0.6760] 0.87869 [0.8141]
Test of restrictions on a fully contemporaneous structure26: =
10.525 [0.1042]
Jo
ur
na
lP
re
-p
ro of
Chi^2(6)
26
The test has been undertaken on a simultaneous system characterized by a fully contemporaneous causal feedback matrix where all equations have been amended so that they all, in addition to the contemporaneous effects of changes to the other (two) endogenous variables, include at least the first lag of their first differences.
45
Appendix C. Macroeconomic effects of permanent shocks
Jo
ur
na
lP
re
-p
ro of
Effects of an increase in the MSCI of 10 %. Deviations from without a financial accelerator (ref. Table 2)
46
47
ro of
-p
re
lP
na
ur
Jo
48
ro of
-p
re
lP
na
ur
Jo
Effects of a 1 percentage point reduction in money market interest rates. Deviations from without a
Jo
ur
na
lP
re
-p
ro of
financial accelerator (ref. Table 4)
49
50
ro of
-p
re
lP
na
ur
Jo
51
ro of
-p
re
lP
na
ur
Jo
Effects of an increase in public expenditure of 1 percent. Deviations from effects without a financial
Jo
ur
na
lP
re
-p
ro of
accelerator (ref. Table 4)
52
53
ro of
-p
re
lP
na
ur
Jo
54
ro of
-p
re
lP
na
ur
Jo
Appendix D. Macroeconomic effects of temporary shocks Effects of an increase in the MSCI of 10 % lasting one year. Deviations from without a financial
Jo
ur
na
lP
re
-p
ro of
accelerator
55
56
ro of
-p
re
lP
na
ur
Jo
57
ro of
-p
re
lP
na
ur
Jo
Effects of a 1 percentage point reduction in the money market interest rate lasting one year. Deviations
Jo
ur
na
lP
re
-p
ro of
from without a financial accelerator
58
59
ro of
-p
re
lP
na
ur
Jo
60
ro of
-p
re
lP
na
ur
Jo
Effects of an increase in public expenditures by 1 percent lasting one year. Deviations from without a
Jo
ur
na
lP
re
-p
ro of
financial accelerator
61
62
ro of
-p
re
lP
na
ur
Jo
63
ro of
-p
re
lP
na
ur
Jo
Appendix E. Macroeconomic effects of a counterfactual shock to oil prices and oil
Jo
ur
na
lP
re
-p
ro of
investments
64
65
ro of
-p
re
lP
na
ur
Jo
66
ro of
-p
re
lP
na
ur
Jo
Appendix F. Capital equations Variables. Source Statistics Norway BORSI = Benchmark stock index Oslo K2NFF = Credit to non-financial corporations in Mainland-Norway PYMN = Deflator GDP Mainland-Norway KAGG = Aggregated Capital BPAK = Average user cost of capital BPA = Average factor price in the industry X = Production L = Employment TIME = Deterministic trend
na
lP
re
-p
ro of
Algebra (example industry 10) A = BORSI/PYMN; a = log(A); Da = diff(a,1); c = log(K2NFF/PYMN); Dc = diff(c,1); kagg10 = log(KAGG10); x10 = log(X10); l10 = log(L10); Dkagg10 = diff(kagg10,1); Dl10= diff(l10,1); Dx10= diff(x10,1); D2c = diff(c,2); Dbpak_Dbpa_10 = Dbpak10-Dbpa10; I:1996(1) = DUM061 = dummy(2006,1,2006,1); = impulse dummy = 1 i 1. quarter 1996, otherwise 0 (The notation I:1996(1), indicates that the impulse dymmy is selected by Autometrics) S1:1996(1) = step dummy = 1 to 1. quarter 1996, thereafter 0 CSeasonal = Seasonal dummy (centered) ECM: defined for each equation in the tables below U = Unrestricted; The automatic selection algorithm in Autometrics is forced to include the variables marked by U
The table for each equation includes standard statistical tests of residual properties, including
ur
autocorrelation (AR), autoregressive conditional heteroscedasticity (ARCH), normality, two additional tests of heteroscedasticity and a specification test (RESET), respectively. The
Jo
residual properties should be considered against the background of experience showing that it is challenging to estimate at such a disaggregated level. In particular, there are problems with heteroscedasticity in the equation for industry 10 (marked with **), heteroscedasticity and AR for industry 63, AR and ARCH for industry 74, AR for industry 81 and heteroscedasticity and specification for industry 85.
67
Table F.1. Industry 10 – Agriculture etc.: ∆kagg10 by OLS: 1991(1)–2013(4)
∆kagg10_1
Coefficient
Std.
t-value
t-prob
Part,R^2
0,3975
0,0863
4,6000
0,0000
0,2055
-0,0181
0,0044
-4,1000
0,0001
0,1698
∆l10
0,0070
0,0042
1,6500
0,1032
0,0321
∆x10
0,0119
0,0032
3,6800
0,0004
0,1417
ECM10_1
-0,0110
0,0029
-3,7800
0,0003
0,1481
DUM061
-0,0078
0,0020
-3,8700
0,0002
0,1547
0,0057
0,0020
2,8700
0,0053
0,0910
CSeasonal
-0,0038
0,0016
-2,4400
0,0167
0,0679
CSeasonal_2
-0,0055
0,0029
-1,9100
0,0601
0,0425
0,0201
0,0057
3,5000
0,0007
0,1302
AR 1-5
F(5,77)
=
0,8028
[0,5511]
ARCH 1-4
F(4,84)
=
0,8305
[0,5095]
Normality
Chi^2(2)
=
1,1419
[0,5650]
Hetero
F(13,77)
=
2,5753
[0,0052]**
Hetero-X
F(38,52)
=
1,5566
[0,0688]
RESET23
F(2,80)
=
0,1917
[0,8260]
Constant
CSeasonal_1
∆2c_1
-p
Sigma
ro of
Residual tests:
0,00194303
re
ECM10 = kagg10-x10+(bpak10-bpa10)+0,006*TIME
Table F.2. Industry 14 – Fishing and hunting: ∆kagg14 by OLS: 1991(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,09263
2,81
0,0061
0,0861
0,09538
2,61
0,0108
0,0749
0,008528
-1,14
0,2584
0,0152
0,002583
-0,858
0,3935
0,0087
0,078368
0,0557
1,41
0,1631
0,023
0,015341
0,009052
1,69
0,0938
0,0331
-0,00881
0,002655
-3,32
0,0013
0,1158
0,039962
0,0109
3,67
0,0004
0,138
F(5,79)
=
0,233
[0,9469]
ARCH 1-4
F(4,84)
=
1,913
[0,1158]
Normality
Chi^2(2)
=
6,3346
[0,0421]*
Hetero
F(11,79)
=
0,75274
[0,6850]
Hetero-X
F(26,64)
=
0,82799
[0,6971]
RESET23
F(2,82)
=
1,6371
[0,2008]
∆kagg14_1
0,260516
∆kagg14_2
0,248821
Constant
-0,0097 -0,00222
na
ECM1413_1 ∆c ∆a_2
ur
CSeasonal DUM081
lP
Coefficient
Residual tests:
Jo
AR 1-5
Sigma
0,010584
ECM1413 = kagg14 -x14 + (bpak14-bpa14)+0,006*TIME
68
Table F.3. Industry 15 – Consumer goods: ∆kagg_ECM_1_15 by OLS: 1991(1)–2013(4) Coefficient
Std.
t-value
t-prob
Part,R^2
∆c_2
0,075541
0,02003
3,77
0,0003
0,1494
CSeasonal
-0,00836
0,001328
-6,3
0
0,3288
CSeasonal_1
-0,00411
0,001272
-3,23
0,0018
0,1139
CSeasonal_2
-0,00595
0,001244
-4,78
0
0,2201
-0,21146
0,00421
-50,2
0
0,9689
I:1997(4)
0,210836
0,004247
49,6
0
0,9682
I:2002(4)
0,012467
0,004209
2,96
0,004
0,0977
I:2005(1)
-0,01053
0,004229
-2,49
0,0148
0,0711
I:2007(2)
-0,01168
0,004288
-2,72
0,0079
0,0839
Constant U
-0,02991
0,000463
-64,5
0
0,9809
∆bpak_∆bpa_15 U
-0,01691
0,008185
-2,07
0,042
0,0501
AR 1-5
F(5,76)
=
2,7311
[0,0253]*
ARCH 1-4
F(4,84)
=
0,33045
[0,8568]
Normality
Chi^2(2)
=
1,29
[0,5247]
F(7,79)
=
1,1753
[0,3265]
Hetero-X
F(14,72)
=
0,73978
[0,7278]
RESET23
F(2,79)
=
1,765
Hetero
0,00410642
[0,1779]
re
Sigma
-p
Residual tests:
ro of
I:1997(2)
ECM15 = kagg15-0,90086*x15+(bpak15-bpa15)
Coefficient 0,454925
CSeasonal
-0,00983
CSeasonal_1
-0,00628
CSeasonal_2 I:1995(1)
S1:2001(1)
t-value
t-prob
Part,R^2
8,01
0
0,4609
0,00083
-11,8
0
0,6517
0,000805
-7,8
0
0,4476
-0,00966
0,000905
-10,7
0
0,6031
0,010907
0,002746
3,97
0,0002
0,1738
0,004262
0,001385
3,08
0,0029
0,1121
0,009153
0,002898
3,16
0,0023
0,1174
ur
S1:1999(4) S1:2000(4)
Std.
0,05681
na
∆kagg25_3
lP
Table F.4. Industry 25 – Intermediate goods etc.: ∆kagg25 by OLS: 1991(1)–2013(4)
-0,01186
0,002702
-4,39
0
0,2042
0,007941
0,001653
4,81
0
0,2354
-0,01381
0,002385
-5,79
0
0,309
S1:2008(4)
0,013677
0,002005
6,82
0
0,3828
Constant U
-0,04804
0,009759
-4,92
0
0,2442
∆c_1 U
0,030863
0,01559
1,98
0,0514
0,0497
∆c_2 U
0,042351
0,01613
2,63
0,0105
0,0842
∆a_3 U
0,004709
0,002407
1,96
0,0541
0,0486
∆a_4 U
0,002938
0,002241
1,31
0,1939
0,0224
ECM25_1 U
-0,00908
0,002029
-4,47
0
0,2107
S1:2007(3)
Jo
S1:2008(2)
69
Residual tests: AR 1-5
F(5,70)
=
2,3578
[0,0490]*
ARCH 1-4
F(4,84)
=
0,42549
[0,7898]
Normality
Chi^2(2)
=
1,3976
[0,4972]
Hetero
F(20,69)
=
0,85255
[0,6438]
Hetero-X
F(53,36)
=
1,0403
[0,4567]
RESET23
F(2,73)
=
1,9803
[0,1454]
Sigma
0,00251241
ECM25 = kagg25-0,90909*x25+(bpak25-bpa25)
Table F.5. Industry 30 – Energy-intensive goods: ∆kagg30 by OLS: 1991(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,775708
0,05625
13,8
0
0,7065
CSeasonal
-0,01348
0,001011
-13,3
0
0,6923
CSeasonal_2
-0,00464
0,000891
-5,21
0
0,2557
I:1995(1)
0,011241
0,00346
3,25
0,0017
0,1179
I:2008(2)
0,017789
0,003434
5,18
0
0,2536
S1:1992(4)
0,013521
0,003613
3,74
0,0003
0,1506
S1:1993(1)
-0,01171
0,003471
-3,37
0,0012
0,126
S1:2005(3)
-0,01283
0,003456
-3,71
0,0004
0,1487
S1:2005(4)
0,029417
0,004834
6,09
0
0,3192
S1:2006(1)
-0,01387
0,003595
-3,86
0,0002
0,1585
Constant U
-0,00952
0,005234
-1,82
0,0727
0,0402
ECM30_1 U
-0,00459
0,002996
-1,53
0,1295
0,0289
∆c U
0,026076
0,01879
1,39
0,169
0,0238
=
1,8842
[0,1073]
=
0,87198
[0,4843]
Chi^2(2)
=
2,3886
[0,3029]
F(10,76)
=
0,72016
[0,7030]
F(19,67)
=
0,51501
[0,9465]
=
1,2138
[0,3027]
F(5,74)
ARCH 1-4
F(4,84)
na
AR 1-5
Normality Hetero
F(2,77)
-p
re
0,00332436
Jo
sigma
ur
Hetero-X RESET23
lP
Residual tests:
ro of
Coefficient ∆kagg30_1
ECM30 = kagg30-0,86957*x30+(bpak30-bpa30);
70
Table F.6. Industry 45 – Engineering products: ∆kagg45 by OLS: 1991(1)–2013(4) Coefficient
Std.
t-value
t-prob
Part,R^2
0,520107
0,09183
5,66
0
0,2764
-0,0073
0,01115
-0,655
0,5144
0,0051
∆l45_1
-0,03743
0,006185
-6,05
0
0,3036
DUM951
0,092098
0,005158
17,9
0
0,7914
DUM952
-0,04493
0,009572
-4,69
0
0,2078
ECM45_1
-0,00155
0,002277
-0,679
0,4989
0,0055
DUM971
0,019092
0,005172
3,69
0,0004
0,1396
∆2c
0,042658
0,01715
2,49
0,0149
0,0686
AR 1-5
F(5,79)
=
1,5409
[0,1867]
ARCH 1-4
F(4,84)
=
0,24789
[0,9102]
Normality
Chi^2(2)
=
3,486
[0,1750]
∆kagg45_1 Constant
F(8,80)
=
2,0718
[0,0483]*
Hetero-X
F(14,74)
=
1,4035
[0,1732]
RESET23
F(2,82)
=
0,40925
[0,6655]
sigma
-p
Hetero
ro of
Residual tests:
0,00503507
re
ECM45 = kagg45-0,97143*x45+(bpak45-bpa45);
Table F.7. Industry 55 – Construction: ∆kagg55 by OLS: 1991(1)–2013(4)
0,742179
∆kagg55_3
0,148245
∆m55
0,035492
Seasonal
0,008451
t-value
I:1993(2)
t-prob
Part,R^2
0,04414
16,8
0
0,7859
0,04247
3,49
0,0008
0,1366
0,00613
5,79
0
0,3033
0,001073
7,88
0
0,4462
na
∆kagg55_1
Std.
lP
Coefficient
0,015646
0,003959
3,95
0,0002
0,1687
-0,02474
0,00395
-6,26
0
0,3376
0,03339
0,00398
8,39
0
0,4776
-0,01904
0,004014
-4,74
0
0,2262
-0,01199
0,003893
-3,08
0,0029
0,1096
0,034848
0,003893
8,95
0
0,51
-0,02222
0,004287
-5,18
0
0,2587
Constant U
-0,03596
0,01298
-2,77
0,007
0,0906
ECM55_1 U
-0,00595
0,00221
-2,69
0,0087
0,086
∆c_1 U
0,040993
0,02147
1,91
0,0599
0,0452
∆a U
0,006481
0,003843
1,69
0,0958
0,0356
I:1995(1)
I:1998(1) I:1999(2) I:2002(1)
Jo
I:2002(2)
ur
I:1996(1)
71
Residual tests: AR 1-5
F(5,72)
=
1,1232
[0,3559]
ARCH 1-4
F(4,84)
=
1,2195
[0,3087]
Normality
Chi^2(2)
=
1,7672
[0,4133]
Hetero
F(13,71)
=
1,2685
[0,2525]
Hetero-X
F(28,56)
=
1,3663
[0,1589]
RESET23
F(2,75)
=
1,7132
[0,1873]
Sigma
0,00377077
ECM55 = kagg55-x55+(bpak55-bpa55)
Table F.8. Industry 63 – Banking and insurance: ∆kagg63 by OLS: 1983(4)–2013(4) Coefficient Std. t-value t-prob Part,R^2 0,06775
8,16
0
∆kagg63_2
0,159941
0,06488
2,47
0,0152
Constant
0,002972
0,000902
3,29
0,0013
∆m63
0,016412
0,005339
3,07
0,0027
ECM63_1
-0,01387
0,003335
-4,16
0,0001
∆c_2
0,034611
0,01832
1,89
0,0615
DUM061
0,032896
0,005134
6,41
0
DUM071
0,026517
0,005091
5,21
CSeasonal
0,004481
0,001557
2,88
CSeasonal_2
0,002745
0,00142
1,93
AR 1-5
F(5,106)
=
ARCH 1-4
F(4,113)
=
Normality
Chi^2(2)
=
Hetero
F(12,106)
=
Hetero-X
F(32,86)
=
RESET23
F(2,109)
=
0,00488742
na
sigma
0,0311 0,27
0,0326
re
0,0557
[0,8630] [0,5869]
2,7124
[0,0031]**
1,746
[0,0221]*
2,5115
[0,0858]
ur
0,1348
0,0695
1,0656
Jo
0,0785
0,0048
0,32165
72
0,0891
0,1964
[0,0015]**
ECM63 = kagg63-0,5*x63+(bpak63-bpa63);
0,0519
0
4,22
lP
Residual tests:
0,375
ro of
0,55289
-p
∆kagg63_1
Table F.9. Industry 71 – Electricity: ∆kagg71 by OLS: 1991(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,53296
0,1006
5,3
0
0,2622
∆kagg71_2
0,321126
0,0996
3,22
0,0018
0,1163
∆kagg71_3
0,172414
0,09963
1,73
0,0875
0,0365
I:1993(1)
0,003854
0,001475
2,61
0,0107
0,0796
I:2009(1)
-0,00507
0,001483
-3,42
0,001
0,1287
I:2013(1)
-0,00431
0,001545
-2,79
0,0067
0,0895
Constant U
0,003612
0,00078
4,63
0
0,2137
DUM124 U
0,00355
0,001467
2,42
0,0178
0,069
Seasonal U
-0,00854
0,000629
-13,6
0
0,6998
ECM71_1 U
-0,00105
0,001096
-0,962
0,339
0,0116
∆c_2 U
0,007765
0,007541
1,03
0,3063
0,0132
Seasonal_1 U
-0,00209
0,000965
-2,17
0,0332
0,0561
Seasonal_2 U
-0,00197
0,001022
-1,93
0,0571
0,0451
AR 1-5
F(5,74)
=
1,7378
[0,1365]
ARCH 1-4
F(4,84)
=
0,45481
[0,7686]
Normality
Chi^2(2)
=
1,2771
[0,5281]
Hetero
F(13,74)
=
0,90865
[0,5482]
Hetero-X
F(23,64)
=
0,94363
[0,5449]
RESET23
F(2,77)
=
0,77103
[0,4661]
re
= kagg71-x71+(bpak71-bpa71)+0,002*TIME
na
ECM71
0,00138158
lP
sigma
-p
Residual tests:
ro of
Coefficient ∆kagg71_1
Table F.10. Industry 74 – Domestic transport: ∆kagg74 by OLS: 1981(1)–2013(4)
DUM901
Std.
t-value
t-prob
Part,R^2
-0,04297
0,01501
-2,86
0,0049
0,0615
ur
Constant
Coefficient
-0,16598
0,007517
-22,1
0
0,7959
0,099179
0,00748
13,3
0
0,5845
DUM001
-0,04646
0,00747
-6,22
0
0,2363
DUM041
0,159784
0,007529
21,2
0
0,7827
ECM74_1
-0,01994
0,006147
-3,24
0,0015
0,0777
∆c_1
0,030515
0,02207
1,38
0,1691
0,0151
Jo
DUM931
73
Residual tests: AR 1-5
F(5,120)
=
12,58
[0,0000]**
ARCH 1-4
F(4,124)
=
5,5883
[0,0004]**
Normality
Chi^2(2)
=
4,6613
[0,0972]
Hetero
F(4,123)
=
1,163
[0,3305]
Hetero-X
F(5,122)
=
1,1293
[0,3485]
RESET23
F(2,123)
=
1,4422
[0,2404]
sigma
0,00743528
ECM74 = kagg74-(1/1,3)*x74+(bpak74-bpa74)
Table F.11. Industry 81 – Merchandising: ∆kagg81 by OLS: 1991(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,419986
0,05768
7,28
0
0,3986
∆kagg81_4
0,109089
0,04386
2,49
0,015
0,0718
Constant
-0,03122
0,0227
-1,38
0,1729
0,0231
Seasonal
-0,01745
0,003162
-5,52
0
0,2756
DUM051
-0,16254
0,01104
-14,7
0
0,7303
ECM81_1
-0,00725
0,004917
-1,47
0,1443
0,0265
∆c_2
0,073577
0,05753
1,28
0,2047
0,02
DUM021
-0,08677
0,01086
-7,99
0
0,4436
DUM052
0,171758
0,01467
11,7
0
0,6314
DUM022
0,096502
0,01201
8,04
0
0,4467
DUM041
-0,04641
0,01099
-4,22
0,0001
0,1822
DUM042
0,045125
0,01115
4,05
0,0001
0,17
F(5,75)
ARCH 1-4
F(4,84)
Normality
Chi^2(2)
Hetero Hetero-X
-p
re 7,7698
[0,0000]**
=
2,4458
[0,0526]
=
1,9772
[0,3721]
F(9,76)
=
2,5993
[0,0114]*
F(15,70)
=
1,9525
[0,0319]*
F(2,78)
=
0,78253
[0,4608]
ur
RESET23
sigma
=
na
AR 1-5
lP
Residual tests:
ro of
Coefficient ∆kagg81_1
0,0105689
Jo
ECM81 = kagg81-(1/1,1)*x81+(bpak81-bpa81)+0,003*TIME
74
Table F.12. Industry 85 – Other private services: ∆kagg85 by OLS:1991(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,007072
0,003124
2,26
0,0266
0,0673
CSeasonal
0,003772
0,000982
3,84
0,0003
0,1721
I:1997(1)
0,017075
0,003455
4,94
0
0,256
I:2008(1)
0,028125
0,003602
7,81
0
0,4619
S1:1998(4)
-0,00907
0,00377
-2,4
0,0188
0,0753
S1:1999(1)
0,015203
0,004138
3,67
0,0005
0,1598
S1:1999(4)
0,012622
0,003924
3,22
0,002
0,1272
S1:2000(1)
-0,01612
0,003656
-4,41
0
0,215
S1:2001(3)
0,005679
0,00206
2,76
0,0074
0,0967
S1:2002(4)
0,016262
0,003754
4,33
0
0,209
S1:2003(1)
-0,01877
0,003884
-4,83
0
0,2475
S1:2004(1)
0,010989
0,002601
4,22
0,0001
0,2009
S1:2004(4)
-0,02779
0,003957
-7,02
0
0,41
S1:2005(1)
0,018241
0,003704
4,93
0
0,2546
S1:2006(4)
-0,01557
0,003768
-4,13
0,0001
0,1939
S1:2007(1)
0,022737
0,003612
6,3
0
0,3582
Constant U
-0,06091
0,01821
-3,34
0,0013
0,1361
∆c U
0,054273
0,01938
2,8
0,0066
0,0995
∆c_1 U
0,022179
0,02015
∆c_2 U
0,034527
0,02053
ECM8584_1 U
-0,01085
0,003116
F(5,66)
ARCH 1-4
F(4,84)
na
AR 1-5
Normality Hetero
RESET23
sigma
-p
re
0,2748
0,0168
0,0971
0,0383
-3,48
0,0009
0,1459
=
0,83213
[0,5315]
=
1,4985
[0,2100]
Chi^2(2)
=
0,031246
[0,9845]
F(18,66)
=
2,0089
[0,0213]*
F(33,51)
=
2,1347
[0,0072]**
F(2,69)
=
6,928
[0,0018]**
ur
Hetero-X
1,1
1,68
lP
Residual tests:
ro of
Coefficient ∆a_2
0,00329812
Jo
ECM8584 = kagg85-(1/1,1)*x85+(bpak85-bpa85)-0,005*TIME
75
Table F.13. Industry 86 – Leasing commercial buildings: ∆kagg86 by OLS: 1984(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,427058
0,04991
8,56
0
0,4131
∆kagg86_2
0,305202
0,05149
5,93
0
0,2525
∆kagg86_4
0,13737
0,03454
3,98
0,0001
0,132
I:1997(1)
0,016682
0,003948
4,23
0,0001
0,1465
I:1998(1)
0,011414
0,003986
2,86
0,0051
0,0731
I:2003(2)
0,040325
0,006307
6,39
0
0,2821
I:2003(3)
0,028506
0,006322
4,51
0
0,1635
I:2006(1)
0,025008
0,003983
6,28
0
0,2749
I:2007(1)
0,018422
0,004113
4,48
0
0,1617
Constant U
0,008958
0,003656
2,45
0,016
0,0546
∆m86 U
0,028074
0,003443
8,15
0
0,39
Seasonal U
0,005576
0,000902
6,18
0
0,2688
ECM86_1 U
-0,00357
0,00138
-2,58
0,0111
0,0604
DUM031 U
-0,10249
0,004001
-25,6
0
0,8632
0
0,1491
0,0296
0,0447
0,00391
-4,27
0,003877
0,001757
2,21
Residual tests: =
F(4,112)
=
Normality
Chi^2(2)
=
Hetero
F(13,98)
Hetero-X
F(28,83)
RESET23
F(2,102)
sigma
0,00379004
2,5713
[0,0313]*
2,5807
[0,0411]*
0,64915
[0,7228]
=
1,8278
[0,0490]*
=
1,8915
[0,0139]*
=
1,6315
[0,2007]
lP
F(5,99)
ARCH 1-4
na
AR 1-5
-p
-0,01669
∆a U
re
DUM021 U
ro of
Coefficient ∆kagg86_1
Jo
ur
ECM86 = kagg86-(1/1,5)*x86+(bpak86-bpa86);
76
PII:
S0939-3625(18)30057-8
DOI:
https://doi.org/10.1016/j.ecosys.2019.100731
Reference:
ECOSYS 100731
To appear in:
Economic Systems
Received Date:
10 October 2017
Revised Date:
15 December 2018
Accepted Date:
13 January 2019
Please cite this article as: Benedictow A, Hammersland R, A financial accelerator in the business sector of a macroeconometric model of a small open economy, Economic Systems (2019), doi: https://doi.org/10.1016/j.ecosys.2019.100731
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier.
A financial accelerator in the business sector of a macroeconometric model of a small open economy Andreas Benedictow a and Roger Hammersland b a
Statistics Norway; E-mail address: [email protected]
b
Statistics Norway; E-mail address: [email protected]
Highlights ·A financial accelerator mechanism operating through investments in the business
ro of
sector has been incorporated into a dynamic macroeconometric model of the Norwegian economy
·In this new and amended model, aggregated credit and equity prices are determined
-p
simultaneously in a system characterized by a two-directional contemporaneous causal link, whereby higher equity prices spur more credit and vice versa The sub-system of credit and equity prices has been designed and estimated with a
re
new procedure of simultaneous structural model design
Combined with a mechanism where credit and asset prices are mutually influenced by
credit-asset price spiral.
Simulations illustrate how the introduction of a financial accelerator significantly
na
lP
real investments in the short run, this creates a financial accelerator amplified by a
reinforces and extends the economic cycles in projections and forecasts.
A permanent increase in global equity prices of 10 percent is calculated to increase
ur
Norwegian asset prices by 10 percent in the short run, gradually decreasing to 5 percent in the long run. As credit and equity prices are only allowed to affect the short-term dynamics of
Jo
capital accumulation, this leads only to a minor short-term increase in real business investments, which gradually peters out in the long run.
Monetary policy gets a markedly stronger effect in the short and medium term, while the impact of fiscal policy is affected to a relatively small degree as it is more remotely linked to financial markets
Abstract 1
We have incorporated a financial accelerator mechanism operating through investments in the business sector in a dynamic macroeconometric model of the Norwegian economy. In this new and amended model aggregated credit and equity prices are determined simultaneously in a system characterized by a two-directional contemporaneous causal link, which has been designed and estimated by a new procedure for simultaneous structural model design. Combined with a mechanism where credit and asset prices are mutually influenced by real investments, this creates a financial accelerator amplified by a credit-asset price spiral. Simulations illustrate how the introduction of a financial accelerator significantly reinforces and extends the economic cycles in projections and forecasts, in particular when confronted
ro of
by a severe shock. Furthermore, monetary policy has a markedly stronger effect in the short and medium term, while the impact of fiscal policy is affected to a relatively small degree as it is more remotely linked to financial markets.
JEL classifications: E1, E32, E44
-p
Keywords: Financial variables and the real economy, The financial accelerator, Business
re
cycles, Structural vector error correction modelling, Impulse response analysis, Forecasting
lP
1. Introduction
The idea that conditions in credit markets could affect business cycles has had broad support in the economic literature for many years; see, for example, Hubbard (1998) and
na
Bernanke et al. (1999). The theory of a financial accelerator postulates a reciprocal relationship between access to credit and fixed investment that helps amplify cyclical
ur
fluctuations, see Bernanke and Gertler (1989). Kiyotaki and Moore (1997) took this further by
Jo
introducing an explicit equity price and credit spiral. There has emerged a substantial empirical literature that largely found support for a relationship between (various indicators of) credit availability and macroeconomic fluctuations, see for example Silvestrini and Zaghini (2015). These works were based largely on the equilibrium models of the real business cycle literature (RBC) (see Kydland and Prescott, 1982; Hartley, 1998). In addition, to some extent there are implemented financial accelerator mechanisms in the so-called new
2
Keynesian DSGE models (see Smets and Wouters, 2008; Christensen and Dib, 2008). However, few attempts have been made to incorporate such a mechanism into structural macroeconometric models. An exception is Hammersland and Træe (2014), where two reciprocal and interacting financial accelerator mechanisms are implemented in a macroeconometric model (Bårdsen and Nymoen, 2009) to study the effect of different types of shocks on the financial stability of the Norwegian economy. This model is, however, highly aggregated, and although it contains financial accelerators with origins in both the
ro of
household and business sectors, the interaction between the real economy and the financial variables happens directly via aggregate production (GDP Mainland Norway) and not, as according to economic theory, via its structural sub-components, household consumption and
-p
business investment, respectively.
This paper documents the estimation and implementation of a financial sub-model in
re
a structural macroeconometric model for the Norwegian economy, KVARTS, partly inspired
lP
by Hammersland and Træe (2014).1 However, our implementation is more disaggregated and theory consistent than in previous studies, 1) in that the financial variables affect investment directly and 2) by taking into account that the effect of changing credit and equity prices on
na
investments can be industry-specific. KVARTS is expanded by a financial sub-model where aggregate credit and equity prices in Norway are determined simultaneously in a system
ur
characterized by a two-directional contemporaneous causal link, designed and estimated with
Jo
the help of a new procedure for simultaneous structural model design (Hammersland, 2017). Moreover, the equations in KVARTS for capital formation in each industry are expanded with aggregate credit and/or Norwegian equity prices. The industry-specific real capital is then aggregated to total capital formation, in which the change from one point in time to the next is 1
KVARTS is developed by the research department of Statistics Norway. A full description of KVARTS is beyond the scope of this paper. See Boug et al. (2013A, Appendix A) for an outline of KVARTS. We also refer to Bowitz and Cappelen (2001), Boug et al. (2006), Boug and Fagereng (2010), Benedictow and Boug (2012), Jansen (2013) and Boug et al. (2013A, 2013B) and Hungnes (2016) for descriptions of the main sub-sections of the model.
3
defined as investment in the same period and included as an explanatory variable in the financial sub-model. Thus, equity prices and credit in this model affect real investments, which in turn affect equity prices and credit, and so on. In the financial sub-model, aggregate credit to the mainland industry in the long term is determined by Norwegian equity prices, represented by the Oslo Børs benchmark index and aggregate business investment in mainland Norway, in the short term also by real interest rates, all deflated by the deflator for GDP Mainland Norway. Norwegian equity prices
ro of
are in the long term determined by international equity prices, represented by the global equity price index Morgan Stanley Capital International World (MSCI), real oil prices and the real interest rate, in the short term also by credit to non-financial corporations in mainland
-p
Norway (henceforth referred to as credit).
In KVARTS, gross fixed capital formation (JK) is divided into two main groups of
re
industries; 1) investments in extraction and pipeline transport, and 2) investment in mainland
lP
Norway.2 Investments in the latter group can be divided further into a) investments in public administration, which is an exogenous variable in the model, b) housing investment, which is determined in a separate sub-model for the housing market, where there is also an accelerator
na
mechanism between credit and investment (see Jansen and Anundsen, 2013), and c) business investments, which is the group of industries directly affected by the financial accelerator
ur
presented in this paper.
Jo
The capital stock is determined in KVARTS by 13 estimated, industry-specific equations. The explanatory variables are production and relative factor prices and other relevant variables, such as employment.3 Norwegian equity prices and credit are only
2
Strictly speaking we could also include investments in shipping as a third group here. However, these investments amounted to just 1 percent of the total gross fixed capital formation in 2014. By comparison, 1) was about 29 percent, while 2) a, b and c represented 30, 21 and 20 percent, respectively. 3 Employment (hours worked) is determined in a factor demand system in the same way as real capital, by production and relative factor prices, and, additionally, technological progress (represented by a deterministic trend).
4
included in the short-term dynamics of each equation, and consequently do not affect capital stock in the long term. This is in line with the Modigliani-Miller theorem (Modigliani and Miller, 1958). We find support for an effect of Norwegian equity prices and/or credit in all equations. Gross investment in each industry appears by definition as the change in capital stock from the period before, adjusted for depreciation. The importance of the financial accelerator for the economy as represented in KVARTS is illustrated by three exogenous shifts and a counterfactual experiment that is
ro of
based on a fairly recent experience related to the Norwegian economy. We start with a shift in the global equity price index, MSCI, which only appears in the financial sub-model. A permanent increase of 10 percent in the MSCI leads to a rise in Norwegian equity prices of 10
-p
percent during the first two quarters, followed by a rapid decline gradually decreasing in strength, and converging towards a long-term effect of approximately 5 percent after 6-7
re
years. As credit and (Norwegian) equity prices are only included in the short-term dynamics
lP
of the capital equations, we only get a short-term increase in business investment at just over 1 percent, which gradually disappears within ten years. We then show the additional effect of changes in the money market rate and public demand, respectively, which is attributable to
na
the financial accelerator in KVARTS. A permanent reduction in the three-month money market rate of 1 percentage point leads to a short-term additional increase in business
ur
investment of 1.4 percent after three years due to the financial accelerator. After 8-10 years
Jo
this effect is gone. Next, a permanent increase in government consumption, investment and employment by 1 percent provides only a marginal additional increase in business investment. The additional effect is weak because there is no direct link in the model from public spending to business investment and financial markets. Finally, we look at a counterfactual shock to oil prices and oil investments where these quantities, instead of
5
following their historical paths4 given by the baseline scenario, are set to follow a random walk, basically implying that we maintain the level of these variables at the beginning of 2013 over the entire simulation period. As the data of our baseline scenario implies a substantial and protracted fall in oil prices as well as oil investments of close to, respectively, 40 and 30 percent at the end of the historical sample period in 2017 and only a gradual increase thereafter, this would amount to a substantial boost to the Norwegian economy. If we look at the additional effects of such a counterfactual shift that are due to the inclusion of a financial
ro of
variable per se, these clearly turn out to bear out the relative importance of the financial accelerator in the wake of shocks, as the additional effect attributed to such a model extension is simulated to give an additional boost of close to 1 percent to GDP.
-p
Our results show that introducing a financial accelerator significantly reinforces and extends the economic cycles in the projections and forecasts in KVARTS. In particular,
re
assumptions about future developments in international equity prices prove important for the
lP
estimated business cycle. Monetary policy gets a markedly stronger effect in the short and medium term, while the impact of fiscal policy is affected to a relatively small degree. As far as the counterfactual experiment is concerned, it clearly contributes to demonstrating the
na
intensive role of the financial accelerator in the propagation of severe shocks. In Section 2, we explain the theoretical background of a financial accelerator and
ur
discuss the estimation results for the financial sub-model. In Section 3 we discuss the
Jo
background for the capital equations in KVARTS, which is the point of connection for the financial sub-model, and the estimation results. In Section 4 we highlight the impact of introducing a financial accelerator into a large macroeconomic model by comparing the effects of shifts in exogenous variables in KVARTS with and without the financial accelerator. Section 5 summarizes and concludes. 4
As this experiment is based on simulating the model from the beginning of 2013 to the end of 2020, the last part of the historic data refers to our quarterly forecasts for the period that goes beyond the period for which we have actual data.
6
2. The financial sub-model 2.1 Procyclicality and the financial accelerator The main point of connection between the real economy and financial markets is the private sectors’ need for the external financing of investments. External financing can be obtained either through an expansion of equity by issuing shares or by increasing foreign capital through borrowing.
ro of
The hypothesis of a financial accelerator assumes that equity prices are procyclical. Increased economic activity will lead to higher equity prices, which in turn can give rise to increased investments, higher production and so on. Such a self-reinforcing mechanism can
-p
be explained by standard economic theory. For example, procyclical behavior in the business sector can be justified by increasing equity prices, causing the market price of capital to
lP
This in turn, increases investments.5
re
increase relative to its replacement cost, the relationship known as Tobin’s Q (Tobin, 1969).
However, what is known as a financial accelerator in the literature is strictly speaking an addition to the classic procyclicality described above and arises from the presence
na
of so-called financial frictions. Financial frictions denote conditions that disrupt the players' behavior in the financial market, and in principle cover all costs associated with financial
ur
transactions, be it fees, taxes, time spent, asymmetric information, etc. (see, for example,
Jo
Brunnermeier et al., 2012, for a literature review). Asymmetric information can for instance cause banks to limit their lending to investors who cannot provide sufficient collateral, so that otherwise profitable investments are not being undertaken. Under such circumstances, previously rationed investors may increase borrowing and realize new investments as higher equity prices boost collateral. This may in turn cause the equity prices to rise further, and so
5
Friedman’s permanent income hypothesis (Friedman, 1957) describes a similar mechanism in the household sector, linked to a positive wealth effect on consumption.
7
on. This illustrates how a financial accelerator with a credit and equity price spiral may reinforce economic cycles. Financial frictions are one possible explanation for deviations from the classical financial theory of market adjustments and the hypothesis of efficient markets launched by Eugene Fama (1965).6 Stiglitz and Weiss (1981) and Stiglitz (1982) argue that asymmetric information in particular poses a significant problem and can explain financial instability as well as financial crises. See Brunnermeier (2001, 2008) for literature surveys on bubbles and
ro of
financial frictions and Jermann and Quadrini (2012) and Hirano and Yanagawa (2016) for more recent studies. The international financial crisis around 2008 intensified the interest in financial frictions and their effects on the real economy. Hall (2009) finds empirical evidence
-p
for the existence of such frictions and the importance for financial markets as well as the real economy. Stiglitz (2010) argues that the effects are significant and that the authorities may
re
advantageously reduce frictions through economic policies. Adrian et. al. (2013) also find
lP
some evidence for financial frictions, although the business sector’s overall access to financing due to the financial crisis was largely maintained because bank lending was to a great extent replaced by bonds. Hammersland and Træe (2014) find clear evidence for
na
financial frictions in the Norwegian economy.
ur
2.2 Empirical results for the financial sub-model
Jo
The financial sub-model in KVARTS consists of two econometric equations where aggregate credit7 and equity prices are determined simultaneously and interactively, and total
6
The field of behavioral finance offers an alternative explanation model, which is not bound by classical economic assumptions about rational actors and efficient markets, but instead has its basis in psychology discipline hypotheses about human behavior. This path is not followed in the present paper (see, for example, Diamond and Vartiainen, 2012, for a discussion). 7 Actual credit, i.e. the market solution: We are not able to distinguish between supply and demand for credit.
8
gross investments in the business sector is included as an explanatory variable.8 This specification will capture the presence of both classical procyclicality and a financial accelerator as described above, but cannot tell them apart. For simplicity, we will refer to all procyclicality arising through the financial sub-model in KVARTS as a “financial accelerator”.9 The financial accelerator is estimated and designed simultaneously with a new procedure for simultaneous structural model design (Hammersland, 2017).10 Based on an
ro of
exact identified general model structure, the final dynamic model is designed and estimated simultaneously using the maximum likelihood method. The first step in this procedure is to identify the long-run solution using the methodology of Johansen (1995). There we found 8
Jo
ur
na
lP
re
-p
A large proportion of non-financial enterprises in mainland Norway is not listed, but the main index on the Oslo Stock Exchange can be an indicator for the development in unlisted companies as well. 9 Two alternative methods were considered for incorporating a financial accelerator in KVARTS. The second was to estimate new equations for investments in every single industry, where the industry-specific investments, aggregate credit and equity prices were estimated simultaneously as a three-dimensional structure. Applying this methodology, however, only two industries (81 and 85) turned out to be suitable for empirical modeling in line with economic theory. It may be that there are conditions in macro that are not captured in the disaggregated figures. Another problem was that we did not have access to industry-specific figures for credit and equity prices. Thus, we concluded that the most appropriate approach is to estimate the financial accelerator at an aggregated level and consequently include aggregate credit and equity when estimating the industry-specific capital equations, as described in this report. 10 To give the reader an idea of what this procedure is all about, we here give a brief account of the steps involved (for a more profound and detailed treatment, the interested reader is referred to Hammersland, 2017). The point of departure in the general case is an n-dimensional conditional reduced form VAR of order k where the conditional set of variables, in addition to including ordinary exogenous variables, deterministic terms and dummies, possibly includes a set of “structural” variables that can later serve the role of auxiliary tools to help with the exact identification of the structural model, both in the long and short run. Starting out with what hopefully constitutes a congruent general unrestricted reduced form model (GUM), the first step then involves reducing the model down to a more parsimonious order and then to undertake the long-run analysis (identification, design and estimation) on this version of the model by resorting to the multidimensional procedure of Johansen (1995). In addition to theory, exogenous and deterministic variables with a structural information content (earlier referred to as “structural” variables) can here be utilized to accomplish exact identification. Given exact identification is accomplished, the next step then involves designing the parsimonious version of the long-run structure by letting theory in conjunction with a test of overidentifying restrictions inform the rest of the long-run design process. Having thus arrived at the long-run structure of the model, one then maps the reduced form of the model over to a structural representation (or form) utilizing some of the “structural” exogenous variables included at the outset as tools of exact identification, possibly in conjunction with restrictions on the long-run feedback structure of the model. It is important in this respect to realize that by utilizing so-called structural exogenous information, or restriction on the long-run feedback structure, we can thus accomplish exact identification without having to resort to procedures imposing non-testable restrictions on either the contemporary feedback matrix or the covariance matrix, both attended with highly contentious and controversial issues given their important role in conditioning the properties of the model. After having designed the long-run structure and exactly identified the dynamic structure, conditional on this, the last and final step of the procedure then involves a fully simultaneous and structural reduction (or design) process where all the structural equations are jointly designed by letting tests of overidentifying restrictions inform the reduction process.
9
support for three cointegrating vectors for credit, equity prices and aggregate investments, respectively. Credit is homogeneous of degree one in equity prices and investment, while the equity price is homogeneous of degree 1 in oil prices and international equity prices, plus an additional interest rate effect. Investment is a function of the relationship between equity prices and the replacement cost of capital, i.e. Tobin’s Q. As investments in KVARTS are already determined by the capital equations, the investment equation in the estimated system is replaced by aggregate investments as determined in KVARTS, i.e. the identity presented in
ro of
equation (7) in Section 3.1. Thus, when we estimate the dynamic simultaneous structure in equation (1), the investment equation is taken out of the system.11 The broad (real) equity price index on the Oslo Stock Exchange (rborsi, hereafter referred to as Norwegian equity
-p
prices) and real credit to non-financial corporations in mainland Norway (rk2nff, hereafter referred to as credit) is precisely identified by assuming that the deviation of credit from its
re
long-term equilibrium, rk2nff-0.77rborsi-0.23jk, where jk is real industry investments, only
lP
helps to explain the structural, dynamic course of credit, while disregarding contemporary effects from changes in international equity prices (msci) in the dynamic relation for real
11
ur
na
credit.12,13,14
Jo
See Appendix B Table B.5 for the econometric specification of the long-run aggregate investment relationship. Confer also footnote 15. 12 All variables are deflated by the GDP deflator for mainland Norway, except MSCI, which is interpreted as an indicator of developments in international financial markets. 13 This is just one of many possible ways of accurately identifying (initializing) the design process. Alternatively, one could for example have chosen to ignore the fact that a discrepancy between one of the endogenous variables and its long-term solution can affect the dynamic course of one of the other endogenous variables in the system. One could also have made use of a priori information on the structural properties of some of the deterministic variables in Dt (structural dummy variables), while imposing alternative restrictions on the system’s Г-matrices, i.e. on the coefficients that capture the effects of exogenous factors and lags of the dependent variable as another option. 14 The long-term structure here is overidentified and represents the result of a long-term analysis where the exact identifiable restrictions imply the absence of oil price/international stock price effects in the credit equation and homogeneity of degree one between real stock price, real oil prices and international equity prices in the longterm solution for equity prices. As in the case of the dynamic structure, this is just one of many ways of accurately identifying the long-term structure.
10
k 1 b12 rk 2nff rk 2nff y , j rborsi t j b21 1 rborsi t j 1
(1)
jk k rpoil x , j Dt msci j 1 RR t j
0 rk 2nff 0.77rborsi 0.23 jk 11 11 11 12 0 22 rborsi 0.5rpoi ln ok 0.5msci 0.04 RR t 1 21 22 23
jk 13 rpoil 33 msci RR t
Other explanatory variables in the system are the real oil price in Norwegian kroner
ro of
(rpoilnokt), the real interest rate (RRt) and deterministic variables (Dt) like a constant, seasonal dummies and dummies for structural breaks. Lower case letters indicate logs and Δ indicates change from the preceding period. Θ, Гx and Гy represent coefficient matrices for the effects
-p
of deterministic variables and dynamics, while b, γ and α represent coefficients that capture effects of the contemporary causal relationship between the model endogenous variables,
re
contemporary dynamic effects of changes in exogenous factors and equilibrium correction.
lP
Because of a general lack of automatic “general to specific” modeling algorithms for structural systems, the structural design and reduction process has been carried out manually on quarterly data from Q1 1991 to Q4 2013. The result of this process is given by (2) and
ur
15
na
(3).15,16
Jo
Actually, the structural design process as described in Hammersland (2017) and summarized in Appendix B was originally undertaken on a three-dimensional structure for the fully simultaneous determination of credit, asset prices and aggregate investment. However, as the investment equation in Appendix B Table B.5 is redundant for the practical implementation of the financial accelerator in KVARTS and the partial model of credit and asset prices in equations (2) and (3) is not only included in the system represented by the equations in Tables B.3- B.5, but also precisely estimated without having to specify the investment equation, it seems safe to assume that aggregate investment is exogenous to the dynamic process determining asset prices and credit, that is, contingent on the long-run structure. This is further substantiated by the χ 2 (5)-test in the Appendix, which fails to reject the lack of a contemporaneous causal link going from investment to either asset prices or credit. However, this does not necessarily mean that the estimated long-run structure of the partial model for asset prices and credit is exogenous for the process driving investment in the long run, which is firmly demonstrated by the fact that there, according to the long-run structure of the system in Table B.3-Table B.5, is a two-way causal link between asset prices and credit on the one hand and investment on the other. This is also why the cointegration analysis referred to in the text was undertaken on a fully simultaneous model for the joint distribution of credit, asset prices and investment in the first place. 16 All estimation and model design in this paper has been made utilizing OxMetrics 7.00 (Doornik, 2007).
11
rk 2nff t 0.04 2 rborsit 0.03jk t 2 0.003RRt 0.24rk 2nff t 4 (2)
0.06rk 2nff 0.77rborsi 0.23 jk t 1
rborsit 2.79rk 2nff t 0.03RRt 0.80mscit 0.27 msci t 1 0.17 mscit 2 0.22 rpoilnok t 0.19 rpoilnok t 1
(3)
ro of
0.46rborsi 0.5rpoilnok 0.5msci 0.04 RR t 1
In equation (2), credit in the long run is determined by Norwegian equity prices and
-p
aggregate investments in fixed capital. Additionally, short-term real interest rates are included in the short term (as well as lags of the dependent variable). Equation (3) shows how
re
Norwegian equity prices can be explained by the real oil price, international equity prices and real interest rates in the short and long term, as well as credit in the short term. See Appendix
lP
B for more detailed estimation results and tests.
Figure 1 illustrates the dynamic projections of the simultaneous structural equation
na
systems (2) and (3) within the estimation period from Q3 2007 to Q4 2013. The fit is
Jo
ur
relatively good throughout this period, even through the financial crisis in autumn 2008.17
The financial sub-model described above is connected to KVARTS via investments, as
determined in the estimated industry-specific equations for capital, where Norwegian equity prices and credit are included as explanatory variables, as explained in the following section. 17
In these dynamic forecasts it has not been necessary to include dummies for the financial crises or other events after the end of the estimation period in the third quarter of 2007. That is, these are the unfettered forecasts of the dynamic structure in (2) and (3) itself.
12
3. Demand for capital 3.1 Modelling of industry investments Industry investments are determined endogenously in KVARTS through 13 industryspecific, estimated equations for real capital. The explanatory variables in these capital equations have traditionally been production and relative factor prices, plus other relevant
ro of
variables such as employment, as shown in
(4) ∆𝑘𝑖,𝑡 = 𝛼𝑖 + ∑5𝑗=1 𝛼𝐾,𝑖,𝑗 ∆𝑘𝑖,𝑡−𝑗 + ∑5𝑗=0 𝛼𝑋,𝑖,𝑗 ∆𝑥𝑖,𝑡−𝑗 + ∑5𝑗=0 𝛼𝑃,𝑖,𝑗 ∆𝑝𝑖,𝑡−𝑗
-p
+ ∑5𝑗=0 𝛼𝑍,𝑖,𝑗 ∆𝑧𝑖,𝑡−𝑗 + 𝛽𝑒𝑐𝑚𝑖,𝑡−1 ,
where ki,t is real capital, xi,t production, pi,t relative factor prices and zi,t represents other
1
lP
solution for real capital is given by
re
relevant variables in industry i, period t. α and β are estimated parameters. The long-term
1
na
(5) 𝑒𝑐𝑚𝑖,𝑡 = 𝑘𝑖,𝑡 + 𝑝𝑖,𝑡 − κ 𝑥𝑖,𝑡 + κ 𝛾𝑖 𝑡𝑟𝑒𝑛𝑑,
ur
where κ is the elasticity of scale and γ captures technological development. These two parameters are estimated in a system, as they are common to the demand of all inputs within
Jo
each industry, see Hungnes (2016). In the present paper, we extend the factor demand relationships (4) by including aggregate credit (c) and equity prices (a) in the short-term dynamics, the latter in addition to the capital cost involved in the long-term relationship through the user cost of capital, so that
13
(6) ∆𝑘𝑖,𝑡 = 𝛼𝑖 + ∑5𝑗=1 𝛼𝑘,𝑖,𝑗 ∆𝑘𝑖,𝑡−𝑗 + ∑5𝑗=0 𝛼𝑥,𝑖,𝑗 ∆𝑥𝑖,𝑡−𝑗 + ∑5𝑗=0 𝛼𝑝,𝑖,𝑗 ∆𝑝𝑖,𝑡−𝑗 + ∑5𝑗=0 𝛼𝑧,𝑖,𝑗 ∆𝑧𝑖,𝑡−𝑗 +∑5𝑗=0 𝛼𝑐,𝑗 ∆𝑐𝑡−𝑗 +∑5𝑗=0 𝛼𝑎,𝑗 ∆𝑎𝑡−𝑗 + 𝛽𝑒𝑐𝑚𝑖,𝑡−1 , where ct is credit to non-financial corporations and at represents the main index on the Oslo Stock Exchange at time t. Note that for credit and equity prices, we use the same aggregate variable in all equations. This is because we do not have data available for these variables at the industry level. This simplification implies assuming that aggregate credit and equity prices are good indicators for all industries, i.e. when the main index on the Oslo stock
ro of
exchange rises and when aggregated credit increases, so does the availability of external financing in general at a disaggregated level.
As credit and equity prices are not included in the long-term relationships in (6), the
-p
level of these variables has in principle no effect in the long run.18 As mentioned, only allowing for direct effects of credit and equity in the short term in the capital equations is in
re
line with the Modigliani-Miller theorem, and implies that in the long term the financial
lP
structure, in other words how companies finance themselves, is not relevant to or dependent on the economic cycle.19 Long-term effects of higher capital prices are, however, partially
na
safeguarded by the user cost of capital.
The investments in industry i in period t, JKi,t, appear as the change in capital stock from the previous period, adjusted for depreciation,
ur
(7) JKit = ∆KAGGi,t - δ*KAGGi,t-1,
Jo
where KAGGi,t is the capital stock and δ is the depreciation rate of capital.
3.2 Empirical results 18
Note that the price of capital is included in the definition of the user cost of capital. However, because there is no link in the model between the user cost of capital and the benchmark index on the Oslo stock exchange, the user cost of capital does not represent a channel for long-term effects between equity prices and the accumulation of capital. 19 Strictly speaking, the Modigliani-Miller theorem applies under relatively strong assumptions of efficient markets and absence of taxes and asymmetric information, and provides little information about adjustments in the short term.
14
The sector-specific equations for capital (6) are estimated by ordinary least squares and reduced using the general to specific methodology (see, e.g., Davidson et al., 1978). It is emphasized that the final estimated relations pass standard statistical tests of serial correlation, heteroscedasticity and normal distribution of the residuals. The estimation of the long-term structure (5) is documented in Hungnes (2016). Conditional on these ecm’s, we reestimate the short short-term dynamics of each industry as shown in (6). Potential path dependency in the reduction process is handled with Autometrics (Doornik, 2009), which is also used to search
ro of
for structural breaks and extreme observations. We have included dummy variables where this has contributed to a theory consistent model specification and/or improved the statistical properties.
-p
Table 1 displays the size of investments in the various industries as a share of total industry investments. In 2014, total industry investments constituted 30.1 percent of total
re
investments in the Norwegian economy and 42.3 percent of investments in mainland Norway.
lP
The table also shows that the capital equations for all industries excluding industry 86 includes credit as an explanatory variable and that equity prices are included in five of the 13
na
equations (see Appendix F for detailed estimation results).20
ur
Jo
The financial sub-model/accelerator can easily be “switched off” by exogenizing credit and Norwegian equity prices. We exploit this in the next section in order to identify the impact of including the financial accelerator in KVARTS.
3.3 Impact of the financial accelerator 20
Whether the data supports including aggregate credit and/or equity prices even in the long term may be an interesting topic for further research.
15
The importance of the financial accelerator is illustrated by three alternative simulations in KVARTS, in addition to a counterfactual experiment where we look at the effect of keeping the level of oil investments and the oil price up over an extended historic period where both quantities were subject to significant downward corrections. First, a permanent increase in the MSCI of 10 percent gives a corresponding increase in Norwegian equity prices during the first two quarters, followed by a relatively rapid decline that gradually decreases in strength, converging towards a long-term increase of almost 5
ro of
percent after 6-7 years. Credit increases markedly during the first years, approaching 4 percent after 6-7 years. Note that fiscal policy and the money market interest rate are exogenously determined, so that the expansionary effects of higher international equity prices
-p
are not offset by tighter economic policies.21 The krone exchange rate, on the other hand, is determined endogenously in the model in all three simulations. Note that this is a partial shift.
re
In reality, one could imagine that increasing international equity prices would go hand
lP
in hand with a global upturn in the business cycle, implying that increased international demand for Norwegian goods and services and increased optimism could push Norwegian investments and equity prices even further. As equity prices and credit are not included
na
directly in the long-term solution of the capital equations, there is only a short-term additional
21
Jo
ur
The rationale for the decision of disregarding endogenous policy responses in the main setup is based on the idea of wanting to purely cultivate the impact of the financial accelerator. However, as is suggested by the simulations in Appendix A, where, in addition to presenting some additional simulations related to the standard case, we look at the case where we have implemented (and switched on) an augmented open economy version of a Taylor rule (Taylor, 1993) in the model, policy rules in general only seem to contribute to moderate the effects of the original shifts studied and to dilute the proper contribution of the financial accelerator. As we primarily want to illustrate the partial effect of changes to some exogenous processes and in this respect to study the role played by the financial accelerator in particular, we have therefore chosen to concentrate on the partial effects of shocks, basing this decision on the absence of policy responses beyond what follows from non-discretionary endogenous behavioral model responses. As far as the need for a fiscal policy response is concerned, in this context it is also important to bear in mind that monetary policy after all is considered as the “first line” defense in coping with economic disturbances. With the relatively modest magnitude of the shocks studied in this paper, an exception granted for the counterfactual experiment in Appendix E, where the lack of a policy response overall probably contributes to overstating the role played by the financial accelerator in promoting shocks, there is therefore reason to assume that the role of fiscal policy would be rather limited. Also, fiscal policy enters the model (KVARTS) in a very complex and detailed way. While this makes it suitable for detailed impact analyses of a number of fiscal instruments, it is also difficult to endogenize in a way that properly takes into account a realistic fiscal policy response.
16
increase in industry investments of 1.1 percent and 2 percent in manufacturing investments, which in both cases disappear within about ten years. GDP Mainland Norway gets an additional increase of 0.2 percent the first year, which decreases slowly towards zero. Unemployment falls rapidly by 0.1 percentage point, but this effect is almost gone after 10 years. The additional effect on traditional exports is very tiny, while imports increase more significantly because investments are relatively import-intensive. Table 2 and Figure 2, together with the graphs in Appendix C, show the effects following from the change in the
ro of
MSCI on a range of macroeconomic variables.22 The graphs in Appendix D also show the effects of a temporary one-year increase in the MSCI index on all the variables listed in Table 2. With the notable exception of credit, they all seem to convey the impression of a very low
-p
degree of persistence in the process of adjustment to temporary shocks.
lP
re
While the international equity price index is a new variable in KVARTS through the financial sub-model, interest rates and fiscal policy are, naturally, important features in earlier
na
versions of the model as well. To identify the importance of the accelerator mechanism for the quantification of the economy’s sensitivity to changes in monetary and fiscal policy, we
ur
perform two times two calculations: For each policy area, we first look at the total effects
Jo
when the financial accelerator is switched off, and then check how the financial accelerator changes this, i.e., the additional effect attributable to the financial accelerator.23
22
Credit and Norwegian equity prices are not deflated in the tables in this report. However, the inflationary effects of the shift in international equity prices, and the additional inflation attributable to the financial accelerator in the interest rates and fiscal policy shifts, are marginal. The real effect is thus very similar to the nominal effect. 23 In addition, Appendix A documents the total effects of the two policy shifts with the financial accelerator switched on, that is compared to the baseline scenario.
17
First, we look at a permanent 1 percentage point reduction of the money market interest rate compared to the baseline scenario. Initially, we keep credit and equity prices exogenous, meaning that the financial accelerator is “switched off”. In this way, we can illustrate KVARTS without a financial accelerator before calculating the additional effect attributable to the financial accelerator when it is switched on. In KVARTS, the money market interest rate affects GDP Mainland Norway through two (close to) equally important channels. First, lower interest rates (relative to international rates) make the krona depreciate.
enterprises.
-p
ro of
That in turn leads to higher import prices and strengthens the competitiveness of Norwegian
re
Thus, exports of goods and services increase. Second, lower interest rates lead to
lP
increasing household consumption and demand for housing, and enterprises in the mainland economy increase investments. Industry investments go up both as a direct result of reduced financing costs and due to increased demand for their products. Employment increases and
na
unemployment falls. Real wages go up slightly after a while. After 7 years, overall industry investments are up by 5.6 percent and GDP Mainland Norway by 1.8 percent, see Table 3.
ur
We then switch on the accelerator mechanism (letting the model determine credit and
Jo
Norwegian stock prices) and find that a permanent 1 percentage point reduction in the threemonth money market rate leads to a short-term additional increase in industry investments of 1.4 percent after two years and in manufacturing investment of 2.4 percent after three years, which can be attributed to the financial accelerator. GDP Mainland Norway is 0.2 percent higher after two years, before the effect gradually diminishes. Thus, while the effects of an
18
increase in international equity prices on the real economy came during the first year, it takes 2-3 years before the maximum effect of the interest rate reduction is achieved. The interest rate is included in the long-term solution of the equity price equation, in the short-term dynamics of the credit equation and in the capital equations via the user cost of capital. Thus, the financial accelerator is engaged, both directly through increased demand for credit and equities and through increased investments due to lower user cost of capital. Furthermore, credit and equity prices are included in the capital equations, which, through
ro of
investments, provide new feedback effects to equity prices and credit, and so on. After about 10 years, the additional effects on the real economy die out. Equity prices are permanently higher because interest rates are included in the long-term solution of (3) and credit is
-p
permanently higher because equity prices are included in the long-term credit relationship in (2). Table 4 and Figure 4, together with the graphs in Appendix C, show the additional effects
re
of the interest rate reduction as a result of incorporating a financial accelerator into KVARTS.
lP
However, as depicted by the graphs pertaining to a temporary one-year interest rate shock in Appendix D, the process of adjustment related to these additional effects are not very persistent for most variables, a notable exception again being the effect on credit.
na
Next, we simulate a permanent increase in government consumption, investment and employment by 1 percent, hereafter called public expenditure for convenience. In this
ur
calculation, the money market rate is also kept unchanged.24 Conversely, if the interest rate
Jo
was determined endogenously, the interest rate relation would “lean against the wind”, responding to increased economic activity by higher interest rates, and thus counteract the expansionary fiscal policy. However, our concern here is how fiscal policy works. First, we increase public expenditure with exogenous credit and equity prices, i.e. with the financial accelerator switched off. Higher public expenditure increases GDP directly by leading to 24
The results related to the effects of an increase in public expenditure when switching on the monetary policy rule in the model are given in Table A.4 and Tab A.5 in Appendix A, respectively. They all seem to convey the impression of a somewhat dampened response to shocks compared to what is the case without a policy rule.
19
higher production in the public administration. That leads to higher demand and thus higher production, even in private industries. Higher production in both public and private sectors increases demand for employment and hence leads to lower unemployment and slightly higher wage growth.
ro of
Household demand increases as a result of higher wages and employment. Consumer prices increase less than nominal wages, resulting in higher real wages. After 10 years, GDP
-p
Mainland Norway is 0.6 percent higher, see Table 5.
re
lP
When we make a new corresponding shift with expansionary fiscal policy, but with the financial accelerator switched on, we find that the accelerator only provides a marginal additional increase in industry investments and GDP, see Table 6 and the graphs in,
na
respectively, Figure 4 and the last part of Appendix C pertaining to the public expenditure shock, for a more detailed documentation. The additional effect is small because the model
ur
has no direct link from public expenditure to financial markets via the equations for equity
Jo
prices and credit, but only an indirect link from increased demand working through the capital equations.25 Looking at the graphs of the additional effect pertaining to a temporary one-year shock to public expenditures – albeit minuscule – reveals a rather substantial degree of persistence in the adjustment process for some of the variables, this time not only for credit.
25
Mazzucato (2015) argues that there can be significant direct effects from public to private investment, including through public/private investment partnerships and so-called bell cow effects.
20
Finally, we look at a counterfactual experiment that is based on a fairly recent experience related to the Norwegian economy and that demonstrates its dependence on oil. In particular, in this context we will look at what would have happened if oil prices and investments, instead of falling precipitously from 2013 onwards and staying low for an extended period of time thereafter, had stayed at their level before the shock took place at the
ro of
end of 2013, beginning of 2014, over the simulation period. As shown in the memo of Table 7, such a shock would amount to a gradual increase in oil prices and oil investments relative to the baseline scenario of close to, respectively, 150 and 20 percent over a four- to five-year
-p
period, before settling down at approximately 80 and 16 percent at the end of the simulation period in 2020. As borne out by Table 7, which gives us the additional effects of the
re
counterfactual shift attributable to the inclusion of a financial accelerator, the relative
lP
importance of the financial accelerator in wake of the shocks can be quite substantial. In Table 7 this additional effect is simulated to amount to an additional boost to GDP of close to 1 percent in 2017 before settling down to about 0,6 percent at the end of the simulation
na
period. This process of enhancement is eventually propagated through a substantial upturn on the Oslo stock exchange and a subsequent strong increase in credit, both rendering possible a
Jo
ur
cyclical upturn in investment and consumption.
5. Conclusion We have estimated and implemented a financial sub-model in a macro-econometric model of the Norwegian economy (KVARTS), which takes into account the pro-cyclical
21
interaction between the real economy and financial markets via industry investments. Our implementation is more theory-consistent than previous studies, as the financial variables affect investments directly and we have taken into account that the effects of changing credit and equity prices on investments can be industry-specific. In the financial sub-model, aggregated credit and Norwegian equity prices are determined simultaneously in a twodimensional structural system characterized by full contemporaneous causal interaction, where industry investments are included as an explanatory variable. Furthermore, aggregate
ro of
credit and Norwegian equity prices are included as explanatory variables in the equations for total capital formation in each industry. The industry-specific real capital is aggregated to total capital formation in the mainland industries, which is included in the dynamics of the
-p
financial sub-model. In this way, equity prices and credit affect investments with feedback effects to equity prices and credit and so on.
re
The impact of the financial accelerator is illustrated by three shifts in exogenous
lP
variables in addition to a counterfactual experiment. A permanent reduction in three-month money market interest rates of 1 percentage point provides a short-term additional increase in industry investments by 1.4 percent, attributable to the financial accelerator. A permanent
na
increase in public expenditure of 1 percent provides just a marginal additional increase in industry investments as a result of the financial accelerator, as there is no direct connection
ur
from public expenditure to the financial accelerator. We also shifted the global equity price
Jo
index MSCI, which is included in both the short-term dynamics and the long-term structure in the equation for Norwegian equity prices. A permanent increase in the MSCI of 10 percent provides a long-term increase in Norwegian equity prices of about 5 percent and a short-term increase of about 1 percent only in industrial investments. As far as the counterfactual experiment is concerned, the additional effect that is attributed to the inclusion of the financial variable is simulated to give an additional boost to GDP of close to 1 percent. This is
22
taken as clear evidence of the intensive role played by a financial accelerator in the propagation of shocks. Including the financial sub-model in KVARTS reinforces and extends economic cycles in projections and forecasts for the Norwegian economy. Moreover, monetary policy gets a more important role, as the effect of a change in interest rates is significantly enhanced. The effects of fiscal policy, on the other hand, are affected to a relatively small extent as a
Jo
ur
na
lP
re
-p
ro of
result of this model extension.
23
References Adrian, T., P. Colla and H. S. Shin (2013): “Which Financial Frictions? Parsing the Evidence from the Financial Crisis of 2007-9”: NBER Macroeconomics Annual 2012, Volume 27 (2013), Acemoglu, D., J. Parker and M. Woodford (red.), p. 159 – 214. Elsevier.
Anundsen, A. K. and E. S. Jansen (2013): “Self-reinforcing effects between housing prices and credit”. Journal of Housing Economics, 22(3), p. 192–212.
ro of
Benedictow, A. and P. Boug (2012): “Trade liberalisation and exchange rate pass-through: the case of textiles and wearing apparels”. Empirical Economics 45, p. 757–788.
Bernanke, B. and M. Gertler (1989): “Agency costs, net worth and business fluctuation”.
-p
American Economic Review 79(1), p. 14–31.
re
Bernanke, B., M. Gertler and S. Gilchrist (1999): “The financial accelerator in a quantitative business cycle framework”. Taylor, J. and M. Woodford (red), Handbook of
lP
Macroeconomics, Vol. 1, Elsevier Science B.V, p. 1341–1393.
Boug, P., Cappelen Å. and A. R. Swensen (2006): “Expectations and regime robustness in
na
price formation: evidence from vector autoregressive models and recursive methods”.
ur
Empirical Economics 31, p. 821–845.
Boug, P. and A. Fagereng (2010): “Exchange rate volatility and export performance: a
Jo
cointegrated VAR approach”. Applied Economics 42, p. 851–864
Boug, P., Å. Cappelen and T. Eika (2013A): “The importance of the distribution sector for exchange rate pass-through in a small open economy: a large scale macroeconometric modeling approach”. Discussion Papers 731, Statistics Norway, Appendix A.
24
Boug, P., Å. Cappelen and T. Eika (2013B): “Exchange rate pass-through in a small open economy: the importance of the distribution sector”. Open Economic Review 24, p. 853 - 879. Bowitz, E., Å. Cappelen (2001): “Modeling income policies: some Norwegian experiences 1973-1993”. Economic Modelling 18, p. 349–379. Brunnermeier, M. K. (2001): “Asset Pricing Under Asymmetric Information: Bubbles, Crashes, Technical Analysis, and Herding”. Oxford University Press.
ro of
Brunnermeier, M. K. (2008): ”Bubbles”, in New Palgrave Dictionary of Economics, ed. by S. N. Durlauf, and L. E. Blume. Palgrave Macmillan.
Brunnermeier, M. K., T. M. Eisenbach and Y. Sannikov (2012): “Macroeconomics with
-p
financial frictions: A survey”. NBER Working Paper No. 18102.
re
Bårdsen, G. and R. Nymoen (2009): “Macroeconometric modeling for policy”. Mills, T. C. and K. Patterson (red.), Palgrave Handbook of Econometrics Vol. 2, 17, p. 851–916.
lP
Palgrave-Macmillan.
Christensen, I. and A. Dib (2008): “The financial accelerator in an estimated New Keynesian
na
model”. Review of Economic Dynamics, Vol 1, No 1, p. 155–178.
ur
Davidson, J. E. H., D. F. Hendry, F. Srba, and S. Yeo (1978): “Econometric Modelling of the Aggregate Time-series Relationship Between Consumers’ Expenditure and Income in the
Jo
United Kingdom”. Economic Journal, 88, 352, p. 661—692.
Diamond, P. and H. Vartiainen (2012): “Behavioral Economics and its Applications”. Princeton University Press.
Doornik, J. A. (2007): “Object-Oriented Matrix Programming Using Ox”. 3rd ed. London: Timberlake Consultants Press and Oxford. 25
Doornik, J. A. (2009): “Autometrics”. Castle, J. L. and N. Shepard (2009): “The methodology and practice of econometrics”, Oxford university press.
Fama, E. (1965): “The behavior of stock market prices”. The Journal of Business, Vol. 38, No. 1, p. 34–105. The University of Chicago Press.
ro of
Friedman, M. (1957): “A theory of the consumption function”. Princeton University Press.
-p
Hall, R. E. (2011): “The High Sensitivity of Economic Activity to Financial Frictions”. The
re
Economic Journal 121(552), p. 351–378.
Hammersland, R. (2017): “The Financial Accelerator and the Real Economy: Evidence Using
lP
a data based procedure of Simultaneous Structural Model Design”. Proceedings ITISE 2017
na
(2), p. 1007- 1034.
Hammersland, R. and C. B. Træe (2014): “The financial accelerator and the real economy: A
ur
small macroeconometric model for Norway with financial frictions”. Economic Modeling,
Jo
vol. 36(C), p. 517–537.
Hartley, J., K. Hoover and K. D. Salyer (1998): “Real business cycles: a reader”. Routledge. Hirano, T. and N. Yanagawa (2016):”Asset Bubbles, Endogenous Growth, and Financial Frictions”. Review of Economic Studies 84 (1), p. 406-443.
26
Hubbard, R.G. (1998): “Capital market imperfections and investment”. Journal of Economic Literature 36, p. 193–225. Hungnes, H. (2016): “Using common factors to identify substitution possibilities in a factor demand system with technological changes”. Discussion papers No. 849, Statistics Norway.
Jansen E. S. (2013): “Wealth effects on consumption in financial crisis: the case of Norway”.
ro of
Empirical Economics 45, p. 873–904.
Jermann, U. and V. Quadrini (2012): “Macroeconomic Effects of Financial Shocks”.
-p
American Economic Review, 102(1), p. 238-271.
Johansen, L. (1995): “Likelihood-Based Inference in Cointegrated Vector Autoregressive
re
Models”. New York. Oxford University Press.
lP
Kiyotaki, N. and J. Moore (1997): “Credit cycles”. Journal of Political Economy,
na
105(2), p. 211–248.
Kydland, F. E. and E. C. Prescott (1982): “Time to build and aggregate fluctuations,”
ur
Econometrica 50(6), p. 1345–1369.
Jo
Mazzucato, M. (2015): Building the Entrepreneurial State: A New Framework for Envisioning and Evaluating a Mission-oriented Public Sector”. Wp 824, Levy Economics Institute.
Modigliani, F. and M. Miller (1958): “The Cost of Capital, Corporation Finance and the Theory of Investment”. American Economic Review 48 (3), p. 261–297.
27
Silvestrini, A. and A. Zaghinib (2015): “Financial shocks and the real economy in a nonlinear world: From theory to estimation”. Journal of Policy Modelling 37(6), p. 915–929.
Smets, F. and R. Wouters (2007): “Shocks and frictions in US business cycles: A Bayesian DSGE approach”. American Economic Review, 97 (3), p. 586–606.
ro of
Stiglitz, J. E. (1982): “The Inefficiency of the Stock Market Equilibrium”. Review of Economic Studies, 49(2), p. 241–261.
-p
Stiglitz, J. E., and A. Weiss (1981): “Credit Rationing in Markets with Imperfect
re
Information”. American Economic Review, 71(3), p. 393–410.
lP
Stiglitz, J. E. (2010): “Freefall”. W. W. Norton & Company, New York.
na
Taylor, John B. (1993). "Discretion versus Policy Rules in Practice" (PDF). CarnegieRochester Conference Series on Public Policy. 39: 195–214.
Tobin, J. (1969): “A general equilibrium approach to monetary theory”. Journal of Money
Jo
ur
Credit and Banking 1, p. 15–29.
28
Jo
ur
na
lP
re
-p
ro of
Figure 1. Credit (rk2nff) and Norwegian equities (rborsi), with corresponding, model-based projections
29
Figure 2. Macroeconomic effects of a permanent increase in the MSCI by 10 percent. Deviations from the baseline scenario in percent (corresponding to Table 2) 10
1.5 1.3 1.0 0.8 0.5 0.3 0.0
8 6 4 2 0 2
3
4
5
6
7
8
9 10
1
2
3
4
5
6
7
8
9 10
Oslo benchmark share index
GDP Mainland Norway
Credit to non-financial industries
Industry investments
ro of
1
Figure 3. Macroeconomic effects of a permanent reduction in the Norwegian money market rate of 1 percentage point of incorporating a financial accelerator. Deviations from the model without financial accelerator in percent (corresponding to Table 4) 1.5 1.3 1.0 0.8 0.5 0.3 0.0
10
6 4
re
2
-p
8
0
1
2
3
4
5
6
7
8
9 10
2
3
4
5
6
7
8
9 10
Oslo benchmark share index
lP
GDP Mainland Norway
1
na
Industry investments
Credit to non-financial industries
Figure 4. Macroeconomic effects of a permanent increase in public expenditure by 1 percent of incorporating a financial accelerator. Deviations from the model without financial accelerator in percent (corresponding to Table 6) 10
ur
1.50 1.25 1.00 0.75 0.50 0.25 0.00
8 6
Jo
4 2 0
1 2 3 4 5 6 7 8 9 10
1
2
3
4
5
6
7
8
9 10
GDP Mainland Norway
Oslo benchmark share index
Industry investments
Credit to non-financial industries
30
Table 1. Investment in each industry as a share of total industry investments (value). Inclusion of credit and/or equity prices as explanatory variables in the short-term dynamics of each equation is indicated by X.
Share
Credit
Industry 10 – Agriculture etc.
3.6
X
Industry 14 – Fishing and hunting
1.2
X
Industry 15 – Consumer goods
3.6
X
Industry 25 – Intermediate goods etc.
3.2
X
Industry 30 – Energy-intensive goods
2.5
X
Equity prices
X
X
ro of
Total industry investments in 2014
Industry 45 – Engineering products Industry 55 – Construction Industry 63 - Banking and insurance Industry 71 – Electricity
-p
Industry 74 – Domestic transport Industry 81 – Merchandising
Industry 86 - Leasing commercial buildings Total industry investments
lP
Industry investments as a share of total investments
re
Industry 85 - Other private services
Jo
ur
na
Industry investments as a share of total investments in Mainland Norway
31
6.1
X
7.1
X
3.6
X
8.8
X
9.3
X
6.5
X
27.9
X
16.5 100 30.1 42.3
X
X X
Table 2. Macroeconomic effects of a permanent increase in the MSCI by 10 percent. Deviations from the baseline scenario in percent
2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.2
0.2
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
Household consumption
0.3
0.3
0.3
0.3
0.4
0.3
0.3
0.3
0.2
0.2
-0.1
-0.1
-0.1
-0.1
-0.1
-0.1
-0.1
-0.1
0.0
0.0
0.5
0.5
0.5
0.4
0.4
0.3
0.3
0.3
0.2
0.2
Industries
1.1
1.1
0.9
0.7
0.5
0.4
0.3
0.2
0.2
0.1
Manufacturing
1.3
2.0
1.6
1.3
0.8
0.6
0.4
0.3
0.2
0.2
Exports traditional goods
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Imports
0.3
0.3
0.3
0.3
0.3
0.3
0.2
0.2
0.2
0.1
Wage
0.0
0.0
0.0
0.1
0.1
0.1
0.1
0.1
0.1
0.1
CPI
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.7
1.9
2.8
3.3
9.5
7.3
6.2
5.6
10.0
10.0
10.0
1
Oslo Børs benchmark index Memo
10.0
Jo
ur
na
lP
MSCI 1 Nominal
-p
Unemployment rate (level) Investment in Mainland Norway
ro of
1
3.6
3.7
3.8
3.8
3.8
3.8
5.3
5.1
4.9
4.9
4.9
4.8
10.0
10.0
10.0
10.0
10.0
re
Year
32
10.0
Table 3. Macroeconomic effects without financial accelerator of a permanent reduction in the Norwegian money market rates by 1 percentage point. Deviations from the baseline scenario in percent 2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.3
0.8
1.1
1.3
1.5
1.7
1.8
1.8
1.8
1.7
Household consumption
0.1
0.7
1.2
1.6
2.0
2.2
2.3
2.3
2.2
2.1
-0.7
-0.8
-1.0
-1.2
-1.3
-1.6
-1.7
-1.7
-1.6
-1.4
0.5
1.6
2.3
3.3
4.2
5.1
5.7
5.7
5.5
5.1
Industries
1.0
3.4
3.8
4.3
4.7
5.3
5.6
5.5
5.2
4.8
Manufacturing
0.6
2.7
3.7
3.9
3.4
3.6
3.8
3.8
3.5
3.3
Exports traditional goods
1.6
1.7
1.9
1.8
1.6
1.4
1.3
1.2
1.1
1.1
Imports
0.1
0.7
1.1
1.4
1.7
2.0
2.2
2.2
2.1
1.9
Wage
0.3
0.6
0.8
1.1
1.4
1.6
1.8
1.9
2.1
2.2
CPI
0.4
0.7
0.8
0.9
1.0
1.0
1.1
1.0
1.0
0.9
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
4.4
4.6
4.2
3.9
3.7
3.6
3.5
3.3
3.2
3.0
0.3
0.7
0.9
1.1
1.2
1.4
1.4
1.5
1.6
1.7
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
-1.0
-1.0
-1.0
1
Oslo Børs benchmark index
-1.0
Jo
ur
na
lP
Memo Money market interest rate (percentage points) 1 Nominal
-p
Unemployment rate (level) Investment in Mainland Norway
33
ro of
1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
-1.0
-1.0
-1.0
-1.0
-1.0
re
Year
-1.0
Table 4. Macroeconomic effects of a permanent reduction in Norwegian money market rates by 1 percentage point of incorporating a financial accelerator. Deviations from the effects without a financial accelerator in percent (ref. Table 3) 2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.1
0.2
0.2
0.2
0.2
0.1
0.1
0.1
0.1
0.1
Household consumption
0.1
0.3
0.3
0.3
0.4
0.3
0.3
0.2
0.2
0.2
Unemployment rate (level) Investment in Mainland Norway
0.0
-0.1
-0.1
-0.1
-0.1
-0.1
-0.1
-0.1
0.0
0.0
0.3
0.6
0.7
0.6
0.5
0.5
0.4
0.3
0.2
0.2
Industries
0.6
1.4
1.3
1.2
1.0
0.8
0.6
0.4
0.3
0.1
Manufacturing
0.9
2.2
2.4
2.2
1.7
1.3
0.9
0.6
0.4
0.3
Exports traditional goods
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Imports
0.1
0.3
0.3
0.3
0.3
0.3
0.3
0.2
0.2
0.1
Wage
0.0
0.0
0.0
0.1
0.1
0.1
0.1
0.1
0.1
0.1
CPI
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.0
0.6
2.0
3.3
4.4
Oslo Børs benchmark index1 1 Nominal
4.4
7.8
8.3
7.6
-p 5.1
5.6
5.8
5.9
6.0
5.9
7.1
6.6
6.3
5.9
5.7
5.5
re
na ur Jo 34
ro of
1
lP
Year
Table 5. Macroeconomic effects without financial accelerator of a permanent increase in public expenditure by 1 percent. Deviations from the baseline scenario in percent 2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.3
0.4
0.4
0.4
0.5
0.5
0.6
0.6
0.6
0.6
Household consumption
0.1
0.3
0.4
0.5
0.6
0.6
0.7
0.7
0.7
0.7
Unemployment rate (level) Investment in Mainland Norway
-0.6
-0.4
-0.4
-0.4
-0.4
-0.5
-0.6
-0.6
-0.6
-0.6
0.3
0.3
0.5
0.6
0.8
1.0
1.1
1.2
1.2
1.2
Industries
0.1
0.3
0.3
0.4
0.5
0.6
0.7
0.8
0.7
0.7
Manufacturing
0.0
0.1
0.1
0.1
0.1
0.1
0.2
0.2
0.2
0.2
Exports traditional goods
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Imports
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.8
0.8
0.8
Wage
0.2
0.2
0.3
0.4
0.5
0.5
0.6
0.7
0.7
0.8
CPI
0.0
0.0
0.0
0.1
0.1
0.1
0.2
0.2
0.2
0.3
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.2
0.2
0.2
0.3
0.3
0.4
0.5
0.5
0.5
0.6
0.6
0.6
0.6
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.0
1.0
1.0
Oslo Børs benchmark index Memo
1.0
Jo
ur
na
lP
Public expenditure 1 Nominal
-p
1
35
ro of
1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.0
1.0
1.0
1.0
1.0
re
Year
1.0
Table 6. Macroeconomic effects of a permanent increase in public expenditure by 1 percent of incorporating a financial accelerator. Deviations from the effects without financial accelerator in percent (ref. Table 5) 2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.00
0.00
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
Household consumption
0.00
0.01
0.01
0.01
0.02
0.02
0.02
0.02
0.02
0.02
Unemployment rate (level) Investment in Mainland Norway
0.00
0.00
0.00
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
0.01
0.02
0.03
0.03
0.04
0.04
0.05
0.05
0.04
0.04
Industries
0.01
0.05
0.06
0.07
0.08
0.09
0.09
0.09
0.09
0.08
Manufacturing
0.03
0.09
0.12
0.14
0.14
0.15
0.15
0.15
0.15
0.14
Exports traditional goods
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Imports
0.00
0.01
0.01
0.01
0.02
0.02
0.02
0.02
0.02
0.02
Wage
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.01
0.01
CPI
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.02
0.08
0.15
0.22
0.05
0.15
0.22
1
0.27
Jo
ur
na
lP
Oslo Børs benchmark index 1 Nominal
-p
ro of
1
0.01
0.01
0.01
0.01
0.01
0.01
0.30
0.37
0.45
0.52
0.60
0.66
0.30
0.34
0.39
0.42
0.46
0.50
re
Year
36
Table 7. Macroeconomic effects of a countercyclical shock to oil prices and oil investments. Deviations from the effects without a financial accelerator in percent 2013
2014
2015
2016
2017
2018
2019
2020
GDP Mainland Norway
0.01
0.05
0.38
1.03
1.10
0.91
0.74
0.61
Household consumption
0.03
0.1
0.85
2.26
2.36
1.97
1.71
1.48
-0.01
-0.03
-0.2
-0.54
-0.7
-0.71
-0.63
-0.45
0.04
0.15
1.04
2.9
3.45
3.08
2.47
1.9
Industries
0.09
0.35
2.48
7.07
8.4
7.13
5.23
3.37
Manufacturing
0.04
0.41
2.57
9.46
13.5
12.26
9.05
5.73
Exports traditional goods
0.00
0.00
-0.01
-0.05
-0.13
-0.20
-0.23
-0.21
Imports
0.23
0.08
0.69
1.91
2.13
1.86
1.6
1.31
Wage
0.00
0.01
0.03
0.11
0.23
0.36
0.48
0.58
-0.00
-0.00
-0.02
-0.07
-0.07
-0.03
0.03
0.08
0.00
-0.00
-0.01
-0.06
-0.08
-0.04
0.01
0.07
0.01
0.03
0.17
0.32
0.49
0.54
0.57
0.55
0.03
0.28
1.98
0.95
4.2
28.08
3.499
13.10
0.41
0.48
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1 1
Oslo Børs benchmark index Memo: Oil price
31.43
109.91
155.48
122.87
106.51
92.39
80.08
5.67
15.87
21.66
19.94
17.09
16.06
Jo
ur
na
lP
Oil investments 1 Nominal
-p
CPI
8.5
17.78
26.04
33.89
65.66
70.83
61.5
50.66
41.42
re
Unemployment rate (level) Investment in Mainland Norway
ro of
Year
37
Appendix A. Macroeconomic effects of economic policy Without a monetary policy rule Table A.1. Macroeconomic effects including a financial accelerator of a permanent reduction in the Norwegian money market rates by 1 percentage point. Deviations from the baseline scenario in percent. År
2
3
4
5
6
7
8
9
10
0.9
1.2
1.5
1.7
1.9
2.0
1.9
1.9
1.7
0.2
1.0
1.5
1.9
2.3
2.5
2.6
2.6
2.4
2.2
-0.7
-0.9
-1.1
-1.3
-1.4
-1.7
-1.8
-1.8
-1.6
-1.4
0.8
2.3
3.0
3.9
4.7
5.6
6.1
6.1
5.8
5.2
Industries
1.6
4.8
5.2
5.6
5.8
6.2
6.3
5.9
5.4
5.0
Manufacturing
1.6
5.0
6.2
6.2
5.2
5.0
4.8
4.4
3.9
3.6
Exports traditional goods
1.6
1.7
1.9
1.8
1.5
1.4
1.2
1.2
1.1
1.1
Imports
0.3
1.0
1.4
1.7
2.0
2.3
2.4
2.4
2.2
2.0
Wage
0.3
0.6
0.9
1.2
1.5
1.7
CPI
0.4
0.7
0.8
0.9
1.0
1.1
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
4.4
4.6
4.2
3.9
3.7
3.6
0.3
0.8
1.0
1.2
1.4
1.5
0.6
2.0
3.5
4.6
5.4
5.9
4.6
8.5
9.0
8.2
-1.0
-1.0
-1.0
-1.0
Unemployment rate (level) Investment in Mainland Norway
1
Oslo Børs benchmark index
1.9
2.0
2.2
2.4
1.1
1.0
1.0
0.9
3.5
3.3
3.2
3.1
1.5
1.6
1.7
1.8
6.2
6.3
6.3
6.3
7.6
7.1
6.7
6.3
6.0
5.8
-1.0
-1.0
-1.0
-1.0
-1.0
-1.0
re
Memo Money market interest rate (percentage points) 1 Nominal.
ro of
Household consumption
1 0.4
-p
GDP Mainland Norway
År GDP Mainland Norway Household consumption
Industries Manufacturing Imports Wage
4
5
6
7
8
9
10
0.5
0.5
0.5
0.6
0.6
0.6
0.6
0.3
0.4
0.5
0.6
0.7
0.7
0.7
0.7
0.7
-0.4
-0.4
-0.4
-0.4
-0.6
-0.6
-0.6
-0.6
-0.6
0.4
0.5
0.7
0.9
1.0
1.2
1.2
1.3
1.2
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.8
0.8
0.8
0.0
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.8
0.8
0.8
0.2
0.2
0.3
0.4
0.5
0.5
0.6
0.7
0.7
0.8
0.0
0.0
0.0
0.1
0.1
0.1
0.2
0.2
0.2
0.3
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.2
0.2
0.2
0.3
0.3
0.4
0.5
0.5
0.5
0.6
0.6
0.6
0.6
0.0
0.1
0.1
0.2
0.3
0.4
0.4
0.5
0.6
0.7
Oslo Børs benchmark index1
0.0
0.2
0.2
0.3
0.3
0.3
0.4
0.4
0.5
0.5
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
Jo
CPI
3
0.4
0.3
ur
Exports traditional goods
0.1 -0.6
2
0.4
na
Unemployment rate (level) Investment in Mainland Norway
1 0.3
lP
Table A.2. Macroeconomic effects including a financial accelerator of a permanent increase in government spending by 1 percent. Deviations from the baseline scenario in percent.
Memo Public expenditure 1 Nominal.
38
With a monetary policy rule Table A.3. Macroeconomic effects of a permanent increase in the MSCI by 10 percent. Deviations from the baseline scenario in percent.
Year
1
2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.16
0.15
0.15
0.13
0.13
0.12
0.1
0.08
0.06
0.05
Household consumption
0.33
0.31
0.29
0.33
0.35
0.31
0.26
0.21
0.17
0.14
-0.09
-0.09
-0.1
-0.10
-0.09
-0.08
-0.06
-0.04
0.02
0.0
0.51
0.5
0.45
0.39
0.33
0.29
0.24
0.18
0.13
0.06
Industries
1.08
1.09
0.89
0.69
0.49
0.32
0.2
0.12
0.07
0.05
Manufacturing
1.29
2.02
1.63
1.22
0.76
0.49
0.28
0.18
0.16
0.16
Unemployment rate (level) Investment in Mainland Norway
0.0
0.0
-0.01
-0.02
-0.02
-0.02
-0.01
-0.01
0.0
0.28
0.28
0.28
0.28
0.28
0.24
0.20
0.16
0.12
0.08
Wage
0.02
0.03
0.04
0.06
0.07
0.08
0.08
0.08
0.07
0.07
CPI
-0.00
-0.01
-0.01
-0.01
-0.0
-0.0
0.00
0.01
0.02
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1 Oslo Børs benchmark index1
-0.00
-0.00
-0.03
-0.05
-0.06
-0.06
-0.06
-0.05
0.01 -0.026
-0.00
0.09
0.13
0.14
0.13
0.13
0.09
0.06
0.05
0.04
0.04
0.73
1.89
2.8
3.32
3.58
3.7
3.74
3.74
3.72
3.71
9.52
7.27
6.20
5.57
5.15
4.91
4.79
4.74
4.74
4.76
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
Jo
ur
na
lP
MSCI 1 Nominal.
re
Memo
39
ro of
0.0
Imports
-p
Exports traditional goods
Table A.4. Macroeconomic effects without financial accelerator of a permanent increase in public expenditure by 1 percent. Deviations from the baseline scenario in percent. Year GDP Mainland Norway Household consumption Unemployment rate (level) Investment in Mainland Norway Industries
1
2
3
4
5
6
7
8
9
10
0.3
0.3
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.1
0.2
0.3
0.4
0.5
0.5
0.5
0.5
0.5
0.5
-0.5
-0.3
-0.3
-0.4
-0.3
-0.4
-0.5
-0.5
-0.5
-0.5
0.3
0.3
0.4
0.5
0.6
0.7
0.8
0.8
0.8
0.7
0.1
0.1
0.1
0.2
0.3
0.3
0.3
0.3
0.2
0.2
0.0
0.0
0.1
0.1
0.1
0.1
0.2
0.2
0.2
0.2
-0.1
-0.1
-0.1
-0.1
-0.1
-0.2
-0.2
-0.2
-0.2
-0.2
Imports
0.2
0.3
0.4
0.5
0.5
0.6
0.6
0.6
0.6
0.6
Wage
0.1
0.2
0.3
0.3
0.4
0.5
0.5
0.5
0.6
0.6
CPI
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.2
-0.1
-0.2
-0.2
-0.2
-0.2
-0.2
-0.3
-0.3
-0.2
-0.2
0.3
0.3
0.4
0.4
0.5
0.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.0
1.0
1.0
1.0
1.0
1.0
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1 Oslo Børs benchmark index1
Jo
ur
na
lP
re
Public expenditure 1 Nominal.
40
0.5
0.5
0.5
0.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.0
1.0
1.0
1.0
-p
Memo
ro of
Manufacturing Exports traditional goods
Table A.5. Macroeconomic effects of a permanent increase in public expenditure by 1 percent of incorporating a financial accelerator. Deviations from the effects without financial accelerator in percent (ref. Table 5 for a comparison).
2
3
4
5
6
7
8
9
10
GDP Mainland Norway
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Household consumption
0.00
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
Unemployment rate (level) Investment in Mainland Norway
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
-0.01
-0.00
0.00
-0.00
-0.00
-0.00
-0.01
-0.00
-0.00
Industries
0.00
-0.02
-0.00
0.00
-0.00
-0.00
-0.00
-0.01
-0.00
0.01
Manufacturing
0.02
0.01
0.01
0.02
0.01
0.01
0.01
0.00
0.01
0.01
Exports traditional goods
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Imports
0.00
-0.01
-0.00
-0.00
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
Wage
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.01
0.01
CPI
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
NOKEUR Household real disposable income excluding dividends Credit non-financial corporations Mainland Norway1
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.02
Oslo Børs benchmark index1
-0.07
-0.21
-0.17
-0.18
1.0
1.0
1.0
1.0
Jo
ur
na
lP
Public expenditure 1 Nominal.
-p
Memo
41
ro of
1
0.02
0.03
0.03
0.03
0.04
0.05
-0.24
-0.27
-0.31
-0.34
-0.33
-0.31
1.0
1.0
1.0
1.0
1.0
re
Year
1.0
Appendix B. Financial sub-model Variables. Source: Statistics Norway, except MSCI: Macrobond. BORSI = Benchmark stock index Oslo (OSEBX) K2NFF = Credit to non-financial corporations in Mainland-Norway JK = I = Gross fixed capital formation REALR = Real interest rate MSCI = Morgan Stanley Capital International: Broad global equity benchmark index, see www.msci.com POILNOK = Oil price, Brent blend PYMN = Deflator GDP Mainland-Norway PJK = Price index for investments / user costs RC= Real repurchasing costs Algebra.
ro of
A = BORSI/PYMN; a = log(A); Da = diff(a,1); D4a = diff(a,4); DDa = diff(Da,1);
-p
RC=PJK/PYMN; rc=log(RC);
re
pymn = log(PYMN); Dpymn= diff(pymn,1);
lP
c = log(K2NFF /PYMN); Dc = diff(c,1);
na
RPOILNOK = POILNOK/PYMN; rpoilnok = log (RPOILNOK); Drpoilnok = diff(rpoilnok,1); DREALR = diff(REALR,1)
ur
i = log(I); Di = diff(i,1);
Jo
msci=log(MSCI); Dmsci =diff(msci,1);
DUMYYQ = Dummy equal to 1 in year YY and quarter Q, zero otherwise. IDUMYYQ = Impuls dummy equal to 1 -1 in respectively quarter Q and Q+1, year YY.
42
Table B.1. Equation for ∆c. CFIML: 1990(4)-2013(4) Coefficient
Std.
t-value
t-prob
CIa_1
-0,05595
0,006704
-8,34
0
Constant
0,635864
0,0755
8,42
0
∆i_2
0,023965
0,01374
1,74
0,0873
∆a
0,036147
0,01056
3,42
0,0012
∆a_1
-0,03615
0,01056
-3,42
0,0012
∆REALR
-0,0031
0,002531
-1,23
0,2263
∆c_4
0,174756
0,06565
2,66
0,0104
DUM933
-0,03055
0,01424
-2,14
0,0368
sigma = 0,0151042
Table B.2. Equation for ∆a. CFIML: 1990(4)-2013(4) Std.
t-value
t-prob
CIb_1
-0,4597
0,06037
-7,61
0
∆c
2,79348
0,5184
5,39
0
Constant
-3,13626
0,4109
-7,63
0
∆REALR
-0,0272
0,01174
-2,32
0,0247
∆msci
0,792869
0,08321
9,53
0
∆msci_1
0,268681
0,07605
3,53
0,0009
∆msci_2
0,166451
0,07735
2,15
0,0362
∆rpoilnok
0,224819
0,05797
3,88
0,0003
∆rpoilnok_1
-0,19894
0,05225
-3,81
0,0004
DUM08Q4
-0,19296
0,06962
-2,77
0,0078
IDUM08Q1
-0,08674
0,0428
-2,03
0,048
re
lP
na
sigma = 0,0694159
-p
Coefficient
ro of
CIa = c-0,770716*a-0,229284*i
CIb = a+0,0351764*REALR-0,493792*msci-0,506208*rpoilnok
ur
BFGS using analytical derivatives (eps1=0.0001; eps2=0.005): Strong convergence log-likelihood 389.109982 no. of observations 93
Jo
Vector SEM-AR 1-5 test: Vector Normality test: Vector Hetero test:
-T/2log|Omega| no. of parameters
F(20,146) = Chi^2(4) = F(216,54) =
43
653.032549 18
1.1069 [0.3486] 3.5719 [0.4670] 0.98715 [0.5411]
Table B.3. Equation for ∆c. CFIML: 1990(4)-2013(4) Coefficient
Std.
t-value
t-prob
CIa_1
-0,05595
0,006704
-8,34
0
Constant
0,635864
0,0755
8,42
0
∆i_2
0,023965
0,01374
1,74
0,0873
∆a
0,036147
0,01056
3,42
0,0012
∆a_1
-0,03615
0,01056
-3,42
0,0012
-0,0031
0,002531
-1,23
0,2263
∆c_4
0,174756
0,06565
2,66
0,0104
DUM933
-0,03055
0,01424
-2,14
0,0368
∆REALR
sigma = 0,0151042
Table B.4. Equation for ∆a. CFIML: 1990(4)-2013(4) Coefficient
Std.
t-value
t-prob
CIb_1
-0,4597
0,06037
-7,61
0
∆c
2,79348
0,5184
5,39
0
-3,13626
0,4109
-7,63
0
Constant ∆REALR
0,01174
-2,32
0,0247
0,792869
0,08321
9,53
0
∆msci_1
0,268681
0,07605
3,53
0,0009
∆msci_2
0,166451
0,07735
2,15
0,0362
∆rpoilnok
0,224819
0,05797
3,88
0,0003
∆rpoilnok_1
-0,19894
0,05225
-3,81
0,0004
DUM08Q4
-0,19296
0,06962
-2,77
0,0078
IDUM08Q1
-0,08674
0,0428
-2,03
0,048
lP
re
-p
-0,0272
∆msci
ro of
CIa = c-0,770716*a-0,229284*i
sigma = 0,0694159
na
CIb = a+0,0351764*REALR-0,493792*msci-0,506208*rpoilnok
Table B.5. Equation for ∆i. CFIML: 1990(4)-2013(4) Coefficient
Std.
t-value
t-prob
0,06821
-4,15
0
3,41037
0,8191
4,16
0
-0,252202
0,09309
-2,71
0,0092
0.186126
0.07937
2.35
0.0229
0,082196
0,05108
1,61
0,1138
--0.050786
0.02846
-1.78
0.0803
CS
-0,122141
0,02436
-5,0
0,0000
CS_1
-0,081953
0,01929
-4,25
0,0001
CS_2
-0,059541
0,01755
-3,29
0,0013
Constant ∆i_1
Jo
∆i_4
-0,283233
ur
CIc_1
∆a
∆4a_1
sigma = 0,0517617 CIc = i-0,60573*(a-rc)
BFGS using analytical derivatives (eps1=0.0001; eps2=0.005): 44
Strong convergence log-likelihood: 540.610624 no. of observations 93 Vector SEM-AR 1-5 test: Vector Normality test: Vector Hetero test:
-T/2log|Omega| no. of parameters
F(45,199) = Chi^2(6) = F(420,110)=
936.494475 27
1.0070 [0.4685] 4.0047 [0.6760] 0.87869 [0.8141]
Test of restrictions on a fully contemporaneous structure26: =
10.525 [0.1042]
Jo
ur
na
lP
re
-p
ro of
Chi^2(6)
26
The test has been undertaken on a simultaneous system characterized by a fully contemporaneous causal feedback matrix where all equations have been amended so that they all, in addition to the contemporaneous effects of changes to the other (two) endogenous variables, include at least the first lag of their first differences.
45
Appendix C. Macroeconomic effects of permanent shocks
Jo
ur
na
lP
re
-p
ro of
Effects of an increase in the MSCI of 10 %. Deviations from without a financial accelerator (ref. Table 2)
46
47
ro of
-p
re
lP
na
ur
Jo
48
ro of
-p
re
lP
na
ur
Jo
Effects of a 1 percentage point reduction in money market interest rates. Deviations from without a
Jo
ur
na
lP
re
-p
ro of
financial accelerator (ref. Table 4)
49
50
ro of
-p
re
lP
na
ur
Jo
51
ro of
-p
re
lP
na
ur
Jo
Effects of an increase in public expenditure of 1 percent. Deviations from effects without a financial
Jo
ur
na
lP
re
-p
ro of
accelerator (ref. Table 4)
52
53
ro of
-p
re
lP
na
ur
Jo
54
ro of
-p
re
lP
na
ur
Jo
Appendix D. Macroeconomic effects of temporary shocks Effects of an increase in the MSCI of 10 % lasting one year. Deviations from without a financial
Jo
ur
na
lP
re
-p
ro of
accelerator
55
56
ro of
-p
re
lP
na
ur
Jo
57
ro of
-p
re
lP
na
ur
Jo
Effects of a 1 percentage point reduction in the money market interest rate lasting one year. Deviations
Jo
ur
na
lP
re
-p
ro of
from without a financial accelerator
58
59
ro of
-p
re
lP
na
ur
Jo
60
ro of
-p
re
lP
na
ur
Jo
Effects of an increase in public expenditures by 1 percent lasting one year. Deviations from without a
Jo
ur
na
lP
re
-p
ro of
financial accelerator
61
62
ro of
-p
re
lP
na
ur
Jo
63
ro of
-p
re
lP
na
ur
Jo
Appendix E. Macroeconomic effects of a counterfactual shock to oil prices and oil
Jo
ur
na
lP
re
-p
ro of
investments
64
65
ro of
-p
re
lP
na
ur
Jo
66
ro of
-p
re
lP
na
ur
Jo
Appendix F. Capital equations Variables. Source Statistics Norway BORSI = Benchmark stock index Oslo K2NFF = Credit to non-financial corporations in Mainland-Norway PYMN = Deflator GDP Mainland-Norway KAGG = Aggregated Capital BPAK = Average user cost of capital BPA = Average factor price in the industry X = Production L = Employment TIME = Deterministic trend
na
lP
re
-p
ro of
Algebra (example industry 10) A = BORSI/PYMN; a = log(A); Da = diff(a,1); c = log(K2NFF/PYMN); Dc = diff(c,1); kagg10 = log(KAGG10); x10 = log(X10); l10 = log(L10); Dkagg10 = diff(kagg10,1); Dl10= diff(l10,1); Dx10= diff(x10,1); D2c = diff(c,2); Dbpak_Dbpa_10 = Dbpak10-Dbpa10; I:1996(1) = DUM061 = dummy(2006,1,2006,1); = impulse dummy = 1 i 1. quarter 1996, otherwise 0 (The notation I:1996(1), indicates that the impulse dymmy is selected by Autometrics) S1:1996(1) = step dummy = 1 to 1. quarter 1996, thereafter 0 CSeasonal = Seasonal dummy (centered) ECM: defined for each equation in the tables below U = Unrestricted; The automatic selection algorithm in Autometrics is forced to include the variables marked by U
The table for each equation includes standard statistical tests of residual properties, including
ur
autocorrelation (AR), autoregressive conditional heteroscedasticity (ARCH), normality, two additional tests of heteroscedasticity and a specification test (RESET), respectively. The
Jo
residual properties should be considered against the background of experience showing that it is challenging to estimate at such a disaggregated level. In particular, there are problems with heteroscedasticity in the equation for industry 10 (marked with **), heteroscedasticity and AR for industry 63, AR and ARCH for industry 74, AR for industry 81 and heteroscedasticity and specification for industry 85.
67
Table F.1. Industry 10 – Agriculture etc.: ∆kagg10 by OLS: 1991(1)–2013(4)
∆kagg10_1
Coefficient
Std.
t-value
t-prob
Part,R^2
0,3975
0,0863
4,6000
0,0000
0,2055
-0,0181
0,0044
-4,1000
0,0001
0,1698
∆l10
0,0070
0,0042
1,6500
0,1032
0,0321
∆x10
0,0119
0,0032
3,6800
0,0004
0,1417
ECM10_1
-0,0110
0,0029
-3,7800
0,0003
0,1481
DUM061
-0,0078
0,0020
-3,8700
0,0002
0,1547
0,0057
0,0020
2,8700
0,0053
0,0910
CSeasonal
-0,0038
0,0016
-2,4400
0,0167
0,0679
CSeasonal_2
-0,0055
0,0029
-1,9100
0,0601
0,0425
0,0201
0,0057
3,5000
0,0007
0,1302
AR 1-5
F(5,77)
=
0,8028
[0,5511]
ARCH 1-4
F(4,84)
=
0,8305
[0,5095]
Normality
Chi^2(2)
=
1,1419
[0,5650]
Hetero
F(13,77)
=
2,5753
[0,0052]**
Hetero-X
F(38,52)
=
1,5566
[0,0688]
RESET23
F(2,80)
=
0,1917
[0,8260]
Constant
CSeasonal_1
∆2c_1
-p
Sigma
ro of
Residual tests:
0,00194303
re
ECM10 = kagg10-x10+(bpak10-bpa10)+0,006*TIME
Table F.2. Industry 14 – Fishing and hunting: ∆kagg14 by OLS: 1991(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,09263
2,81
0,0061
0,0861
0,09538
2,61
0,0108
0,0749
0,008528
-1,14
0,2584
0,0152
0,002583
-0,858
0,3935
0,0087
0,078368
0,0557
1,41
0,1631
0,023
0,015341
0,009052
1,69
0,0938
0,0331
-0,00881
0,002655
-3,32
0,0013
0,1158
0,039962
0,0109
3,67
0,0004
0,138
F(5,79)
=
0,233
[0,9469]
ARCH 1-4
F(4,84)
=
1,913
[0,1158]
Normality
Chi^2(2)
=
6,3346
[0,0421]*
Hetero
F(11,79)
=
0,75274
[0,6850]
Hetero-X
F(26,64)
=
0,82799
[0,6971]
RESET23
F(2,82)
=
1,6371
[0,2008]
∆kagg14_1
0,260516
∆kagg14_2
0,248821
Constant
-0,0097 -0,00222
na
ECM1413_1 ∆c ∆a_2
ur
CSeasonal DUM081
lP
Coefficient
Residual tests:
Jo
AR 1-5
Sigma
0,010584
ECM1413 = kagg14 -x14 + (bpak14-bpa14)+0,006*TIME
68
Table F.3. Industry 15 – Consumer goods: ∆kagg_ECM_1_15 by OLS: 1991(1)–2013(4) Coefficient
Std.
t-value
t-prob
Part,R^2
∆c_2
0,075541
0,02003
3,77
0,0003
0,1494
CSeasonal
-0,00836
0,001328
-6,3
0
0,3288
CSeasonal_1
-0,00411
0,001272
-3,23
0,0018
0,1139
CSeasonal_2
-0,00595
0,001244
-4,78
0
0,2201
-0,21146
0,00421
-50,2
0
0,9689
I:1997(4)
0,210836
0,004247
49,6
0
0,9682
I:2002(4)
0,012467
0,004209
2,96
0,004
0,0977
I:2005(1)
-0,01053
0,004229
-2,49
0,0148
0,0711
I:2007(2)
-0,01168
0,004288
-2,72
0,0079
0,0839
Constant U
-0,02991
0,000463
-64,5
0
0,9809
∆bpak_∆bpa_15 U
-0,01691
0,008185
-2,07
0,042
0,0501
AR 1-5
F(5,76)
=
2,7311
[0,0253]*
ARCH 1-4
F(4,84)
=
0,33045
[0,8568]
Normality
Chi^2(2)
=
1,29
[0,5247]
F(7,79)
=
1,1753
[0,3265]
Hetero-X
F(14,72)
=
0,73978
[0,7278]
RESET23
F(2,79)
=
1,765
Hetero
0,00410642
[0,1779]
re
Sigma
-p
Residual tests:
ro of
I:1997(2)
ECM15 = kagg15-0,90086*x15+(bpak15-bpa15)
Coefficient 0,454925
CSeasonal
-0,00983
CSeasonal_1
-0,00628
CSeasonal_2 I:1995(1)
S1:2001(1)
t-value
t-prob
Part,R^2
8,01
0
0,4609
0,00083
-11,8
0
0,6517
0,000805
-7,8
0
0,4476
-0,00966
0,000905
-10,7
0
0,6031
0,010907
0,002746
3,97
0,0002
0,1738
0,004262
0,001385
3,08
0,0029
0,1121
0,009153
0,002898
3,16
0,0023
0,1174
ur
S1:1999(4) S1:2000(4)
Std.
0,05681
na
∆kagg25_3
lP
Table F.4. Industry 25 – Intermediate goods etc.: ∆kagg25 by OLS: 1991(1)–2013(4)
-0,01186
0,002702
-4,39
0
0,2042
0,007941
0,001653
4,81
0
0,2354
-0,01381
0,002385
-5,79
0
0,309
S1:2008(4)
0,013677
0,002005
6,82
0
0,3828
Constant U
-0,04804
0,009759
-4,92
0
0,2442
∆c_1 U
0,030863
0,01559
1,98
0,0514
0,0497
∆c_2 U
0,042351
0,01613
2,63
0,0105
0,0842
∆a_3 U
0,004709
0,002407
1,96
0,0541
0,0486
∆a_4 U
0,002938
0,002241
1,31
0,1939
0,0224
ECM25_1 U
-0,00908
0,002029
-4,47
0
0,2107
S1:2007(3)
Jo
S1:2008(2)
69
Residual tests: AR 1-5
F(5,70)
=
2,3578
[0,0490]*
ARCH 1-4
F(4,84)
=
0,42549
[0,7898]
Normality
Chi^2(2)
=
1,3976
[0,4972]
Hetero
F(20,69)
=
0,85255
[0,6438]
Hetero-X
F(53,36)
=
1,0403
[0,4567]
RESET23
F(2,73)
=
1,9803
[0,1454]
Sigma
0,00251241
ECM25 = kagg25-0,90909*x25+(bpak25-bpa25)
Table F.5. Industry 30 – Energy-intensive goods: ∆kagg30 by OLS: 1991(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,775708
0,05625
13,8
0
0,7065
CSeasonal
-0,01348
0,001011
-13,3
0
0,6923
CSeasonal_2
-0,00464
0,000891
-5,21
0
0,2557
I:1995(1)
0,011241
0,00346
3,25
0,0017
0,1179
I:2008(2)
0,017789
0,003434
5,18
0
0,2536
S1:1992(4)
0,013521
0,003613
3,74
0,0003
0,1506
S1:1993(1)
-0,01171
0,003471
-3,37
0,0012
0,126
S1:2005(3)
-0,01283
0,003456
-3,71
0,0004
0,1487
S1:2005(4)
0,029417
0,004834
6,09
0
0,3192
S1:2006(1)
-0,01387
0,003595
-3,86
0,0002
0,1585
Constant U
-0,00952
0,005234
-1,82
0,0727
0,0402
ECM30_1 U
-0,00459
0,002996
-1,53
0,1295
0,0289
∆c U
0,026076
0,01879
1,39
0,169
0,0238
=
1,8842
[0,1073]
=
0,87198
[0,4843]
Chi^2(2)
=
2,3886
[0,3029]
F(10,76)
=
0,72016
[0,7030]
F(19,67)
=
0,51501
[0,9465]
=
1,2138
[0,3027]
F(5,74)
ARCH 1-4
F(4,84)
na
AR 1-5
Normality Hetero
F(2,77)
-p
re
0,00332436
Jo
sigma
ur
Hetero-X RESET23
lP
Residual tests:
ro of
Coefficient ∆kagg30_1
ECM30 = kagg30-0,86957*x30+(bpak30-bpa30);
70
Table F.6. Industry 45 – Engineering products: ∆kagg45 by OLS: 1991(1)–2013(4) Coefficient
Std.
t-value
t-prob
Part,R^2
0,520107
0,09183
5,66
0
0,2764
-0,0073
0,01115
-0,655
0,5144
0,0051
∆l45_1
-0,03743
0,006185
-6,05
0
0,3036
DUM951
0,092098
0,005158
17,9
0
0,7914
DUM952
-0,04493
0,009572
-4,69
0
0,2078
ECM45_1
-0,00155
0,002277
-0,679
0,4989
0,0055
DUM971
0,019092
0,005172
3,69
0,0004
0,1396
∆2c
0,042658
0,01715
2,49
0,0149
0,0686
AR 1-5
F(5,79)
=
1,5409
[0,1867]
ARCH 1-4
F(4,84)
=
0,24789
[0,9102]
Normality
Chi^2(2)
=
3,486
[0,1750]
∆kagg45_1 Constant
F(8,80)
=
2,0718
[0,0483]*
Hetero-X
F(14,74)
=
1,4035
[0,1732]
RESET23
F(2,82)
=
0,40925
[0,6655]
sigma
-p
Hetero
ro of
Residual tests:
0,00503507
re
ECM45 = kagg45-0,97143*x45+(bpak45-bpa45);
Table F.7. Industry 55 – Construction: ∆kagg55 by OLS: 1991(1)–2013(4)
0,742179
∆kagg55_3
0,148245
∆m55
0,035492
Seasonal
0,008451
t-value
I:1993(2)
t-prob
Part,R^2
0,04414
16,8
0
0,7859
0,04247
3,49
0,0008
0,1366
0,00613
5,79
0
0,3033
0,001073
7,88
0
0,4462
na
∆kagg55_1
Std.
lP
Coefficient
0,015646
0,003959
3,95
0,0002
0,1687
-0,02474
0,00395
-6,26
0
0,3376
0,03339
0,00398
8,39
0
0,4776
-0,01904
0,004014
-4,74
0
0,2262
-0,01199
0,003893
-3,08
0,0029
0,1096
0,034848
0,003893
8,95
0
0,51
-0,02222
0,004287
-5,18
0
0,2587
Constant U
-0,03596
0,01298
-2,77
0,007
0,0906
ECM55_1 U
-0,00595
0,00221
-2,69
0,0087
0,086
∆c_1 U
0,040993
0,02147
1,91
0,0599
0,0452
∆a U
0,006481
0,003843
1,69
0,0958
0,0356
I:1995(1)
I:1998(1) I:1999(2) I:2002(1)
Jo
I:2002(2)
ur
I:1996(1)
71
Residual tests: AR 1-5
F(5,72)
=
1,1232
[0,3559]
ARCH 1-4
F(4,84)
=
1,2195
[0,3087]
Normality
Chi^2(2)
=
1,7672
[0,4133]
Hetero
F(13,71)
=
1,2685
[0,2525]
Hetero-X
F(28,56)
=
1,3663
[0,1589]
RESET23
F(2,75)
=
1,7132
[0,1873]
Sigma
0,00377077
ECM55 = kagg55-x55+(bpak55-bpa55)
Table F.8. Industry 63 – Banking and insurance: ∆kagg63 by OLS: 1983(4)–2013(4) Coefficient Std. t-value t-prob Part,R^2 0,06775
8,16
0
∆kagg63_2
0,159941
0,06488
2,47
0,0152
Constant
0,002972
0,000902
3,29
0,0013
∆m63
0,016412
0,005339
3,07
0,0027
ECM63_1
-0,01387
0,003335
-4,16
0,0001
∆c_2
0,034611
0,01832
1,89
0,0615
DUM061
0,032896
0,005134
6,41
0
DUM071
0,026517
0,005091
5,21
CSeasonal
0,004481
0,001557
2,88
CSeasonal_2
0,002745
0,00142
1,93
AR 1-5
F(5,106)
=
ARCH 1-4
F(4,113)
=
Normality
Chi^2(2)
=
Hetero
F(12,106)
=
Hetero-X
F(32,86)
=
RESET23
F(2,109)
=
0,00488742
na
sigma
0,0311 0,27
0,0326
re
0,0557
[0,8630] [0,5869]
2,7124
[0,0031]**
1,746
[0,0221]*
2,5115
[0,0858]
ur
0,1348
0,0695
1,0656
Jo
0,0785
0,0048
0,32165
72
0,0891
0,1964
[0,0015]**
ECM63 = kagg63-0,5*x63+(bpak63-bpa63);
0,0519
0
4,22
lP
Residual tests:
0,375
ro of
0,55289
-p
∆kagg63_1
Table F.9. Industry 71 – Electricity: ∆kagg71 by OLS: 1991(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,53296
0,1006
5,3
0
0,2622
∆kagg71_2
0,321126
0,0996
3,22
0,0018
0,1163
∆kagg71_3
0,172414
0,09963
1,73
0,0875
0,0365
I:1993(1)
0,003854
0,001475
2,61
0,0107
0,0796
I:2009(1)
-0,00507
0,001483
-3,42
0,001
0,1287
I:2013(1)
-0,00431
0,001545
-2,79
0,0067
0,0895
Constant U
0,003612
0,00078
4,63
0
0,2137
DUM124 U
0,00355
0,001467
2,42
0,0178
0,069
Seasonal U
-0,00854
0,000629
-13,6
0
0,6998
ECM71_1 U
-0,00105
0,001096
-0,962
0,339
0,0116
∆c_2 U
0,007765
0,007541
1,03
0,3063
0,0132
Seasonal_1 U
-0,00209
0,000965
-2,17
0,0332
0,0561
Seasonal_2 U
-0,00197
0,001022
-1,93
0,0571
0,0451
AR 1-5
F(5,74)
=
1,7378
[0,1365]
ARCH 1-4
F(4,84)
=
0,45481
[0,7686]
Normality
Chi^2(2)
=
1,2771
[0,5281]
Hetero
F(13,74)
=
0,90865
[0,5482]
Hetero-X
F(23,64)
=
0,94363
[0,5449]
RESET23
F(2,77)
=
0,77103
[0,4661]
re
= kagg71-x71+(bpak71-bpa71)+0,002*TIME
na
ECM71
0,00138158
lP
sigma
-p
Residual tests:
ro of
Coefficient ∆kagg71_1
Table F.10. Industry 74 – Domestic transport: ∆kagg74 by OLS: 1981(1)–2013(4)
DUM901
Std.
t-value
t-prob
Part,R^2
-0,04297
0,01501
-2,86
0,0049
0,0615
ur
Constant
Coefficient
-0,16598
0,007517
-22,1
0
0,7959
0,099179
0,00748
13,3
0
0,5845
DUM001
-0,04646
0,00747
-6,22
0
0,2363
DUM041
0,159784
0,007529
21,2
0
0,7827
ECM74_1
-0,01994
0,006147
-3,24
0,0015
0,0777
∆c_1
0,030515
0,02207
1,38
0,1691
0,0151
Jo
DUM931
73
Residual tests: AR 1-5
F(5,120)
=
12,58
[0,0000]**
ARCH 1-4
F(4,124)
=
5,5883
[0,0004]**
Normality
Chi^2(2)
=
4,6613
[0,0972]
Hetero
F(4,123)
=
1,163
[0,3305]
Hetero-X
F(5,122)
=
1,1293
[0,3485]
RESET23
F(2,123)
=
1,4422
[0,2404]
sigma
0,00743528
ECM74 = kagg74-(1/1,3)*x74+(bpak74-bpa74)
Table F.11. Industry 81 – Merchandising: ∆kagg81 by OLS: 1991(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,419986
0,05768
7,28
0
0,3986
∆kagg81_4
0,109089
0,04386
2,49
0,015
0,0718
Constant
-0,03122
0,0227
-1,38
0,1729
0,0231
Seasonal
-0,01745
0,003162
-5,52
0
0,2756
DUM051
-0,16254
0,01104
-14,7
0
0,7303
ECM81_1
-0,00725
0,004917
-1,47
0,1443
0,0265
∆c_2
0,073577
0,05753
1,28
0,2047
0,02
DUM021
-0,08677
0,01086
-7,99
0
0,4436
DUM052
0,171758
0,01467
11,7
0
0,6314
DUM022
0,096502
0,01201
8,04
0
0,4467
DUM041
-0,04641
0,01099
-4,22
0,0001
0,1822
DUM042
0,045125
0,01115
4,05
0,0001
0,17
F(5,75)
ARCH 1-4
F(4,84)
Normality
Chi^2(2)
Hetero Hetero-X
-p
re 7,7698
[0,0000]**
=
2,4458
[0,0526]
=
1,9772
[0,3721]
F(9,76)
=
2,5993
[0,0114]*
F(15,70)
=
1,9525
[0,0319]*
F(2,78)
=
0,78253
[0,4608]
ur
RESET23
sigma
=
na
AR 1-5
lP
Residual tests:
ro of
Coefficient ∆kagg81_1
0,0105689
Jo
ECM81 = kagg81-(1/1,1)*x81+(bpak81-bpa81)+0,003*TIME
74
Table F.12. Industry 85 – Other private services: ∆kagg85 by OLS:1991(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,007072
0,003124
2,26
0,0266
0,0673
CSeasonal
0,003772
0,000982
3,84
0,0003
0,1721
I:1997(1)
0,017075
0,003455
4,94
0
0,256
I:2008(1)
0,028125
0,003602
7,81
0
0,4619
S1:1998(4)
-0,00907
0,00377
-2,4
0,0188
0,0753
S1:1999(1)
0,015203
0,004138
3,67
0,0005
0,1598
S1:1999(4)
0,012622
0,003924
3,22
0,002
0,1272
S1:2000(1)
-0,01612
0,003656
-4,41
0
0,215
S1:2001(3)
0,005679
0,00206
2,76
0,0074
0,0967
S1:2002(4)
0,016262
0,003754
4,33
0
0,209
S1:2003(1)
-0,01877
0,003884
-4,83
0
0,2475
S1:2004(1)
0,010989
0,002601
4,22
0,0001
0,2009
S1:2004(4)
-0,02779
0,003957
-7,02
0
0,41
S1:2005(1)
0,018241
0,003704
4,93
0
0,2546
S1:2006(4)
-0,01557
0,003768
-4,13
0,0001
0,1939
S1:2007(1)
0,022737
0,003612
6,3
0
0,3582
Constant U
-0,06091
0,01821
-3,34
0,0013
0,1361
∆c U
0,054273
0,01938
2,8
0,0066
0,0995
∆c_1 U
0,022179
0,02015
∆c_2 U
0,034527
0,02053
ECM8584_1 U
-0,01085
0,003116
F(5,66)
ARCH 1-4
F(4,84)
na
AR 1-5
Normality Hetero
RESET23
sigma
-p
re
0,2748
0,0168
0,0971
0,0383
-3,48
0,0009
0,1459
=
0,83213
[0,5315]
=
1,4985
[0,2100]
Chi^2(2)
=
0,031246
[0,9845]
F(18,66)
=
2,0089
[0,0213]*
F(33,51)
=
2,1347
[0,0072]**
F(2,69)
=
6,928
[0,0018]**
ur
Hetero-X
1,1
1,68
lP
Residual tests:
ro of
Coefficient ∆a_2
0,00329812
Jo
ECM8584 = kagg85-(1/1,1)*x85+(bpak85-bpa85)-0,005*TIME
75
Table F.13. Industry 86 – Leasing commercial buildings: ∆kagg86 by OLS: 1984(1)–2013(4) Std.
t-value
t-prob
Part,R^2
0,427058
0,04991
8,56
0
0,4131
∆kagg86_2
0,305202
0,05149
5,93
0
0,2525
∆kagg86_4
0,13737
0,03454
3,98
0,0001
0,132
I:1997(1)
0,016682
0,003948
4,23
0,0001
0,1465
I:1998(1)
0,011414
0,003986
2,86
0,0051
0,0731
I:2003(2)
0,040325
0,006307
6,39
0
0,2821
I:2003(3)
0,028506
0,006322
4,51
0
0,1635
I:2006(1)
0,025008
0,003983
6,28
0
0,2749
I:2007(1)
0,018422
0,004113
4,48
0
0,1617
Constant U
0,008958
0,003656
2,45
0,016
0,0546
∆m86 U
0,028074
0,003443
8,15
0
0,39
Seasonal U
0,005576
0,000902
6,18
0
0,2688
ECM86_1 U
-0,00357
0,00138
-2,58
0,0111
0,0604
DUM031 U
-0,10249
0,004001
-25,6
0
0,8632
0
0,1491
0,0296
0,0447
0,00391
-4,27
0,003877
0,001757
2,21
Residual tests: =
F(4,112)
=
Normality
Chi^2(2)
=
Hetero
F(13,98)
Hetero-X
F(28,83)
RESET23
F(2,102)
sigma
0,00379004
2,5713
[0,0313]*
2,5807
[0,0411]*
0,64915
[0,7228]
=
1,8278
[0,0490]*
=
1,8915
[0,0139]*
=
1,6315
[0,2007]
lP
F(5,99)
ARCH 1-4
na
AR 1-5
-p
-0,01669
∆a U
re
DUM021 U
ro of
Coefficient ∆kagg86_1
Jo
ur
ECM86 = kagg86-(1/1,5)*x86+(bpak86-bpa86);
76