Journal Issue
Share
Article

Russian Federation: Selected Issues Paper

Author(s):
International Monetary Fund
Published Date:
September 2011
Share
  • ShareShare
Show Summary Details

III. A Financial Conditions Index for Russia1

This chapter presents new Financial Conditions Index (FCI) for Russia. By providing a meaningful aggregate measure of monetary and financial conditions, the FCI is an analytical tool that can potentially contribute to the understanding of macro-financial linkages in the Russian economy. The FCI indicates that after a sharp rebound from the nadir of the recession, overall financial conditions in Russia have become moderately positive, thereby providing some support for the recovery. However, conditions have fluctuated considerably over the past year and in the first months of 2011 the FCI declined owing mostly to weakness in house prices.

A. Introduction

1. This chapter presents a Financial Conditions Index (FCI) for Russia. The objective of the FCI is to provide an aggregate measure of monetary and financial conditions in the Russian economy and to analyze their impact on macroeconomic performance. Relative to the more widely-used Monetary Conditions Indices (MCI), the FCI considers a broader set of financial variables that may impact aggregate demand conditions and that may capture important additional channels in the transmission of monetary policy.

2. The FCI is an analytical tool that can potentially contribute to our understanding of macro-financial linkages in the Russian economy. For instance, the FCI provides information on credit conditions and on the relative level of asset prices and how these affect the economy at large. Few existing indicators provide such information, and at present relatively little is known about these relationships in Russia. FCIs are sometimes also used as an input into macroeconomic forecasting models. As a new measure, the results of the Russian FCI need to be interpreted with some caution. The FCI is not intended as an operational tool, and even less so as an operational target. To inform policy decisions, the information from the FCI would need to be complemented with that from broader set of key macroeconomic indicators (for example, monetary policy should be focused on developments in projected inflation, something that is not directly addressed by the FCI).

B. Methodology

3. The FCI aggregates financial variables weighed by their statistical impact on GDP. To construct the FCI, a selection of financial variables is analyzed with respect to each variable’s impact on real GDP. The most significant variables are selected and aggregated with their respective weights determined by their statistical contribution to developments in real GDP. Note that although FCI’s can be used as a leading indicator for economic growth, being a precise forecaster of GDP is not its primary objective. Rather, the measure aims at saying something about financial conditions and the degree to which they support economic activity. Thus, the FCI excludes nonfinancial variables, such as the oil price, even though their inclusion might improve the overall fit of the model.

4. To analyze the impact of the variables on GDP, a vector autoregressive model (VAR) is employed. There are various possible approaches to constructing FCIs.2 In this chapter, following Goodhart and Hofmann (2001) and Swiston (2008), among others, a VAR is used to estimate the coefficients of the FCI. The key advantage of the VAR approach is that it allows for the incorporation of the complex, endogenous two-way interactions between the financial variables and economic activity, and between the financial variables themselves. The FCI weights are determined using impulse response functions from the VAR system.

C. Constructing the FCI for Russia

5. Any statistical analysis of Russian data faces significant challenges which qualify estimation results. The first challenge—owing to the relatively short history of the Russian Federation—is the availability of only very short time series data, with few continuous data series going back beyond the mid-1990s and many financial data series starting only in the 2000s. Also, two major financial crises (1998, 2008–09) during the past 15 years make that segments of the data can be very volatile. The period further coincides with Russia’s structural transformation from an economy in which the financial sector plays a relatively small role, to one in which financial conditions are having an increasing impact on economic performance.

6. Alleviating the constraint of short time series, monthly data are used for a range of financial variables. This approach is facilitated by the fact that the Russian Ministry of Economy produces monthly GDP estimates that are consistent with the official national accounts data. The financial variables that are tested in the different specifications are the (i) real effective exchange rate of the ruble (REER), (ii) the real U.S. 10 year t-bill rate, (iii) the real 3-month MOSPRIME interest rate, (iv) the real average deposit interest rate, (v) real broad money, (vi) the real RTS and MICEX equity indices, (vii) real house prices (existing homes), and (viii) the Russian Economic Barometer (REB) index of borrowing conditions.3 All data are in logs, with the exception of the U.S. and Russian interest rates.

7. Data availability limits the final sample to 2002–10, but the benefits of a longer sample would likely have been modest. The data with respect to house prices, broad money, the MICEX, and the MOSPRIME rate impose constraints on the sample size as they are only available from 2000–02 onwards. Dropping these variables would allow the data sample to start around end-1996. However, experimentation with various specifications of the model suggests that estimations of longer time-periods do not yield more significant results than those based on shorter samples (possibly this finding reflects the increase in the relative impact of financial conditions over time, rendering more recent periods more significant). Against this background, in its final specification, the model is estimated based on a relatively short sample of 2002–10 data, with no constraints on the availability of variables.

8. Data were analyzed both in first derivatives and in deviations from equilibrium (“gaps”). Data were tested in two distinct forms. First, data were analyzed in simple annual percent changes, related to GDP growth. And second, data were tested as deviations from their equilibrium values (or their long-term trend), related to the output gap. The latter approach should be more meaningful since for many financial variables (e.g., interest rates), their relative or absolute levels may matter more for economic growth than changes at the margin.

9. To facilitate the gap approach, equilibrium values were determined for each of the variables. For GDP, the trend—or “potential” growth—is estimated using a modified LRX filter. For the Russian short-term interest rate, consistent with theory, the same trend GDP growth is also used as a proxy for the equilibrium rate, but using a higher smoothing parameter so as to reduce the impact of short-term fluctuations in potential growth. For the REER and house prices, linear trends are calculated, while the equilibrium for the REB credit conditions index is approximated by its all-time historic average. Time-varying trends for the remaining variables were estimated using standard HP filters.

10. In line with expectations, the gap approach was found to yield more significant results, and the findings discussed below are based on this approach. The broad money variable proved insignificant in all model specifications and was therefore eliminated.4 For the domestic interest rate and the stock market index, the best performing measures (MICEX and MOSPRIME, respectively) were selected at the expense of competing similar measures.

D. Estimation Results and FCI

11. Impulse response functions show that the selected variables have a significant impact on the output gap. The chart on the following page shows the results from impulse-response functions (IRF) with respect to the output gap for the remaining variables in the final VAR system. Each of the variables is shown to have a significant impact on the output gap (Y), and with the anticipated, correct sign. The sole exception is the house price variable (HPE), which does have the correct sign but just fails to be significant at a 95 percent confidence interval. House prices are nonetheless maintained in the system because of their theoretical importance as a key financial indicator, and because the variable typically turned significant in alternative specifications of the model. The IRFs illustrate how each of the variables relate to GDP.

Accumulated Response to Cholesky One S.D. Innovations ± 2 S.E.

  • Higher levels of the US long-term interest rate (I_US) and the Russian money market rate (I_RUS) have a clear negative effect on GDP growth, although the effect becomes insignificant relatively quickly for the U.S. interest rate.
  • Broader credit conditions, as measured by the REB index of borrowing conditions (CC), have a particularly strong and significant effect on GDP, suggesting that this measure contains important additional information with respect to credit conditions over and above that contained in interest rates (in particular, information on credit worthiness of borrowers and selectivity of lenders).
  • Asset price increases are also shown to have a positive impact on economic activity, with the impact of share prices (EQ) materializing somewhat faster than that of house prices (HP). Regarding house prices, it should be noted that in the Russian context their impact on economic activity is less likely to come about via the household consumption channel (as, for instance, in the U.S.) since the retail mortgage market is less developed in Russia. Rather, house prices appear to feed through mainly via housing construction activity.

12. Using the accumulated responses of the variables to determine the weights of the individual variables, an FCI is constructed for Russia. Although the maximum (significant) accumulated impact of each of the variables is reached with somewhat different lags, we base the weights on the accumulated response after 12 periods balancing considerations of size and significance of impact across the 6 variables. The resulting FCI is shown on the following page, together with a chart depicting a decomposition of the FCI into its individual components.

13. The FCI performs well and provides an insight into the role of financial conditions in recent cyclical developments. The estimated FCI tracks the output gap well, with a relatively high correlation of 0.75. The charts show how financial indicators loosened appreciably during 2006–08, in particular on the back of low (negative) real interest rates and sharply rising home prices. Conditions deteriorated dramatically when the global crisis hit in late 2008. A sharp tightening of credit conditions—owing to higher real interest rates and, likely, increased differentiation by lenders—was the main driver of this deterioration, compounded by the deep slump in the stock market. From the first quarter of 2009, financial conditions improved sharply, in particular on account of a reduction in real interest rates and easing overall credit conditions. These improvements in financial conditions led the recovery in the real sector.

14. Following the strong rebound from the crisis, more recently overall financial conditions have weakened somewhat, owing to disappointing house price developments. Following a temporary setback in the summer of 2010—owing to the international market upheaval surrounding the financial problems in Greece—the FCI continued improve in the second half of 2010. More recently, however, conditions have tightened somewhat as renewed weakness in the housing market weighed on the FCI. Against this background, financial conditions currently provide only moderate support for the recovery.

FCI versus Output Gap (3-month moving averages)

Decomposition of FCI--2002M1-2010M12

E. The FCI Versus the MCI

15. The FCI is better placed to explain the recent boom and bust than a traditional MCI. To gauge the extent of additional information offered by the FCI over a simpler Monetary Conditions Index (MCI), the chart below shows our new FCI next to the MCI that has been used by the Fund staff as an analytical tool in recent years, and that is based on developments in real interest rates and the REER only. The comparison is illustrative.

FCI versus MCI

(2002-06 average = 100)

  • It is clear from the chart that the MCI provides relatively little insight in the conditions that led to the boom years during 2006–08. Indeed, “monetary conditions,” as measured by the MCI actually tightened for much of this period (owing to an appreciating ruble and rising real interest rates). In contrast, the FCI reveals—by virtue of the inclusion of the REB credit survey and house prices—that broader financial conditions loosened significantly during these years, arguably contributing to the rapid GDP growth and eventual overheating.
  • The FCI also appears to capture better the dire conditions during the recession in 2008/09, when sharply tightening credit standards and plummeting stock prices made financial conditions particularly tight. The MCI, while also falling, shows a considerably less dramatic picture during this period (including because in this index the depreciation of the ruble compensates for much of the spike in real interest rates). Finally, looking at current conditions, the MCI shows a sharp, continuous improvement during 2010, bringing monetary conditions broadly back to pre-crisis levels. The FCI, by contrast, suggests that at the same time overall financial conditions have improved much less, owing to a drag from falling housing prices.
References

    AkerliAhmet AnnaZadornova and JonathanPinder (2010) Financial Conditions Signal Growth Divergence in the New marketsGoldman Sachs New Market Analyst No. 10/01.

    • Search Google Scholar
    • Export Citation

    BeatonKimberly RenéLalonde and CorinneLuu (2009) A Financial Conditions Index for the United States,Bank of Canada Discussion paper 2009–11.

    • Search Google Scholar
    • Export Citation

    GoodhartCharles and BorisHofmann (2001) Asset Prices, Financial Conditions, and the Transmission of Monetary Policy,Paper prepared forthe conference “Asset Prices, Exchange rates, and Monetary Policy,” Stanford University2–3 March.

    • Search Google Scholar
    • Export Citation

    HatziusJan PeterHooper FredericMishkin KermitSchoenholtz and MarkWatson (2010) Financial Conditions Indexes: A Fresh Look after the Financial Crisis,NBER Working Paper No. w16150.

    • Search Google Scholar
    • Export Citation

    SwistonAndrew (2008) A U.S. Financial Conditions Index: Putting Credit Where Credit is Due,IMF Working Paper WP/08/161.

1

Prepared by David Hofman.

2

See Beaton et al. (2009) and Hatzius et al (2010) for useful overviews of the FCI literature and existing FCIs.

3

The REB index of borrowing conditions is based on a survey and measures the share of Russian companies that report an improvement in their borrowing conditions over the preceding month.

4

This result contrasts with that of Akerli et al. (2010) at Goldman Sachs, who find a significant relationship between broad money and output growth and include broad money as one of the variables in their FCI for Russia.

Other Resources Citing This Publication