Business cycles and shocks to financial stability: empirical evidence from a new set of Danish quarterly national accounts 1948–2010

Abstract

In Denmark, consistent time series of quarterly national accounts are only available for the period since 1971. Based on a range of quarterly business cycle indicators, this article constructs a set of non-seasonally adjusted quarterly national accounts for Denmark 1948–1971 using the ‘Proportional Denton Least Square Method’. This compilation method is recommended in the International Monetary Fund's manual on quarterly national accounts as the preferred method to compile quarterly national accounts data consistent with an existing set of annual national accounts. Furthermore, the article reviews some of the empirical evidence on the business cycle and the linkages between the financial sector and the real economy in the Danish post-war period that can be uncovered from the new data-set.

Keywords:

• quarterly national accounts
• Danish economic history
• business cycles
• monetary transmission
• financial stability
• band-pass filters
• VAR analysis

JEL Classification:

• C32
• C82
• E01
• E32
• E44
• E52
• N14

Acknowledgements

The author is grateful to the editors of this journal and two anonymous referees, whose insightful and constructive comments and suggestions have been significant in helping to shape and improve the article in the review process. Furthermore, the author wishes to thank colleagues from Danmarks Nationalbank and Heino Bohn Nielsen, University of Copenhagen, for useful comments on preliminary versions of the results presented in this article. Views and conclusions expressed in the article are those of the author and do not necessarily represent those of Danmarks Nationalbank. The author alone is responsible for any remaining errors.

Notes

¹Cf. Berner/Thage, ‘Kvartalsvise’ (1989).

²Cf. Sørensen, ‘Kvartalsvise’ (1994).

³Cf. Christensen, ‘Kvartalsvise’ (1989).

⁴Thygesen, Sources (1971).

⁵Hansen/Paldam, SMEC (1973).

⁶Ibid.

⁷It is thus left to the users to compile seasonally adjusted data, if needed. This can for example be done via the X-12-ARIMA procedure from the U.S. Census Bureau, cf. U.S. Census Bureau, X-12-ARIMA (2009).

⁸Cf. StatBank Denmark on Statistics Denmark's website (www.dst.dk).

⁹Cf. Christensen, ‘Kvartalsvise’ (1989).

¹⁰Cf. Statistics Denmark, Sources (2008).

¹²Thygesen, Sources (1971).

¹¹The sources for the annual national accounts are Statistics Denmark, ‘Nationalregnskabsstatistik’ (1966) and various issues of Statistiske Efterretninger and the Statistics 10-year review published by Statistics Denmark.

¹³Cf. Denton, ‘Adjustment’ (1971).

¹⁴Bloem et al. Quarterly (2001).

¹⁵Cf. Boot et al. ‘Further Methods’ (1967).

¹⁶Thygesen, Sources (1971).

²³Thygesen, Sources (1971).

²⁴Mordhorst, Dansk (1968).

²⁵Mikkelsen, Dansk (1993), Pedersen, ‘FINDAN’ (1989).

²⁶Statistics Denmark, ‘Kreditmarkedsstatistik’ (1969).

²⁷Busk-Nielsen, Dansk (1996).

²⁸Økonomiministeriet, Danske (1994).

¹⁷Abildgren, ‘Business Cycles’ (2010).

¹⁸Cf. Abildgren, ‘Interest-Rate’ (2005).

¹⁹Abildgren, ‘Business Cycles’ (2010). The data sets presented in the present article have been revised and updated compared with the data sets listed in Abildgren, ‘Business Cycles’ (2010). [

²⁰The website of Danmarks Nationalbank (www.nationalbanken.dk).

²¹Cf. e.g. Stock/Watson, ‘Business Cycle’ (1999) and Agresti/Mojon, ‘Stylised facts’ (2003).

²²Hansen/Knudsen, ‘Korrelationsmønstre’ (2004).

²⁹Baxter/King, ‘Measuring’ (1999). Introductions to filtering methods are found in e.g. Mills, Modelling (2003) and Gencay et al., Wavelets (2002).

³⁰The reason for six quarters as the lower limit (and not zero) is the wish to exclude seasonality and very short-term random fluctuations from the business cycle component. According to the NBER US business cycles have on average been around five years for the post-1854 period and a little more than six years in the post-1970 period. Hansen and Knudsen, ‘Korrelationsmønstre’ (2004) indicate – using both the Baxter-King filter and the Hodrick-Prescott filter – that the post-1974 business cycles in Denmark have been somewhat longer. An upper limit of eight years (32 quarters) therefore seems suitable. The Baxter-King filter ensures that the filtered time series becomes de-trended and stationary in order to avoid spurious cycles. Furthermore, since the filter coefficients are symmetrical, the filtered series have no phase shifts compared with the input series. The filter applied below will be based on a symmetric moving average with 12 observations on each side, which is usually applied in the literature for quarterly data. By transforming a trended input series by natural logarithms before filtering, the cyclical component extracted from the data can (when multiplied by 100) be interpreted as the deviation from the trend measured in percent. This facilitates the economic interpretation of the filtered time series data. In this section, all the time series – except interest rates, the unemployment rate, price- and wage-inflation rates and the banks’ write-down ratio - have been transformed by natural logarithms before filtering. Finally, it should be noted that due to the compilation method of the cyclical components there is a loss of 12 observations at the beginning and at the end of the filtered time series. Since the focus in this section is solely on a historical analysis of the business cycle no forecasts and backcasts have been added to the series in order to compensate for this loss of observations.

³¹Cf. Kydland/Prescott, ‘Business Cycles’ (1990).

³²Cf. King/Watson, ‘Money’ (1996).

³³Cf. e.g. Huang ‘Cyclical Behavior’ (2004).

³⁴Cf. Abildgren, ‘Credit’ (2009).

³⁵Cf. e.g. Blanchard/Simon, ‘Output Volatility’ (2001) and Stock/Watson, ‘Business Cycle’ (2003).

³⁶Cf. Abildgren, ‘Estimates’ (2006).

³⁷Cf. Sims, ‘Money’ (1972), Sims, ‘Macroeconomics’ (1980) and Sims, ‘Comparison’ (1980). A non-technical introduction to VAR models is offered by Stock/Watson, ‘Vector’ (2001).

³⁸Cf. e.g. Marcucci/Quagliariello, ‘Bank portfolio’ (2008), Dovern et al., ‘How resilient’ (2010), Berrospide/Edge, ‘Effects’ (2010) and Österholm, ‘Swedish’ (2010). Studies with an economic-historical perspective include Anari et al., ‘Bank Asset’ (2005) and Kupiec/Ramirez, ‘Bank Failures’ (2009).

³⁹Sløk, ‘Transmissionsmekanismen’ (1997).

⁴⁰Beier/Storgaard, ‘Identifying’ (2006).

⁴¹Recent reviews of the evidence from VAR analysis on the monetary transmission mechanism in the USA and the euro area are found in Christiano et al., ‘Monetary’ (1999) and Peersman/Smets, ‘Monetary’ (2003).

⁴²Cf. Tobin, ‘General’ (1969).

⁴³Cf. Bordo et al., ‘Crisis’ (2001) and Reinhart/Rogoff, ‘Aftermath’ (2009).

⁴⁴Cf. Abildgren et al., ‘Real Economic’ (2011).

⁴⁵Cf. Christensen et al., ‘Nordic’ (1995), Abildgren, ‘Financial’ (2008) and Larsen et al., ‘Landbrug’ (2010).

⁴⁶Hall/McDermott, ‘Quarterly’ (2009), Hayes/Turner, ‘Estimates’ (2007) and Mitchell et al., ‘Monthly’ (2009).

⁴⁷Sims et al., ‘Inference’ (1990).

⁴⁸All econometric results have been obtained via the use of PCGive and JMulTi. The econometrics of VAR-models is covered by Hamilton, ‘Time’ (1994).

⁴⁹The various information criteria applied in this article are described in e.g. Lütkepohl/Krätzig, ‘Applied’ (2004).

⁵⁰Sims et al., ‘Inference’ (1990).

⁵¹Cf. Abildgren, ‘Business Cycles’ (2010). It could be useful to explore the robustness of the findings within the framework of cointegrated VAR models. However, such an analysis is beyond the scope of this article.

⁵²Cf. Abildgren, ‘Business Cycles’ (2010).

⁵³Cf. Abildgren, ‘Business Cycles’ (2010).

⁵⁴For a summary description of Halls method to calculate bootstrapped confidence intervals for impulse responses in VAR models, cf. Lütkepohl/Krätzig, ‘Applied’ (2004), 177–179.