Bosnia and Herzegovina: Selected Economic Issues

This Selected Economic Issues paper for Bosnia and Herzegovina reports that output, exports, and incomes have increased and inflation has stabilized. New modern banking laws have been passed in both entities, and the banking sector has been almost completely privatized, with the majority of assets now under foreign ownership. The reforms to the central bank and to the banking system have been aimed to secure stability and to build an efficient financial system.

Abstract

This Selected Economic Issues paper for Bosnia and Herzegovina reports that output, exports, and incomes have increased and inflation has stabilized. New modern banking laws have been passed in both entities, and the banking sector has been almost completely privatized, with the majority of assets now under foreign ownership. The reforms to the central bank and to the banking system have been aimed to secure stability and to build an efficient financial system.

V. Non-Observed Activity1

A. Introduction

1. Official national accounts data in Bosnia and Herzegovina (BiH) understate activity. A variety of indicators suggest this—including various ratios to GDP such as credit growth (up 21 percentage points of GDP in three years), the velocity of various monetary aggregates (low by regional and international standards), and the external deficit which the central bank estimates at some 23 percent of GDP in 2004. This assessment is widely shared 2

2. The question is “by how much”? This chapter situates estimates for the non-observed activity (NOA) in Bosnia and Herzegovina in an international context, including those for other transition countries. It also summarizes new estimates prepared for BiH suggesting that its NOA is larger than the average in other transition economies, low-middle income, and conflict and post-conflict countries, and is similar to estimates for the CIS area.

B. International Treatment of Non-Observed Activity

3. Official GDP data in all countries reflect the level of activity and the exhaustiveness of data compilation. Taking account of the different concepts involved in unobserved activity, many countries attempt to estimate and integrate into their official GDP data legal activities that are unrecorded (Box 1). Most commonly, statistical offices use direct survey methods—of the labor force, unemployment, household budgets, and businesses—alongside official tax, and licensing records, to gauge this activity. These “direct” methods underlie adjustments to data yielded by standard statistical returns to determine the finally published official GDP data. Sometimes, but less often, corrections are also made for deficiencies in the statistical registries and non-response from the surveyed businesses and households. Only a few countries integrate estimates of illegal activity in published GDP numbers.

Non Observed Activity—Concepts

OECD identifies several elements of NOA: (i) non-response of surveyed businesses or households; (ii) business registries that are not updated; (iii) omission of businesses in the statistical registries; (iv) underreporting of sales and revenues; (v) under registration of legal entities, e.g. for tax evasion; (vi) informal activity on so modest a scale that it is not captured; and (vii) illegal activity.

4. Countries in transition make significant adjustments for NOA. On average the NOA adjustments incorporated in official GDP data amount to 21 percent of officially published GDP. These ratios are some 26 percent in former Soviet Republics (CIS) and 13 percent in Southeastern European countries (SEE) (Text Table 1). This contrasts with developed economies which tend to make only minor adjustments to correct for deliberately unregistered activities and the informal sector, while transition economies attempt to estimate the whole array of non-observed activities. Many transition economies have developed sophisticated survey-based methods to underlie their adjustments for this, but the Former Yugoslav Republics have yet to match this sophistication. Despite acknowledging that NOA may be considerable, the actual adjustments for NOA in Croatia and Serbia and Montenegro amount to just nine percent of published GDP. Only Albania makes a sizeable adjustment, with 27 percent of published GDP coming from this source.

Text Table 1.

GDP Adjustment for NOA in Transition Economies

(In percent of Officially Published GDP)

article image
Source: UN (2003); IMF Country Report No. 05/90; Dell’Anno and Piirisild (2004).

From the Albanian authorities.

Excluding Bosnia and Herzegovina.

Belgium, Canada, Ireland, Italy, and USA.

5. Analysts—but typically not official statisticians—also attempt to gauge NOA using “indirect” methods.3 These typically use various indicators to infer what “true” GDP is (Box 2).

Two Indirect Methods

  • A straightforward method presumes a stable relationship between a variable— perhaps money or electricity consumption—and “true” GDP. An unusually large quantum of currency in circulation or electricity consumption relative to historical norms provides a crude estimator of “increased” NOA activity. The risk in this approach is that factors other than grey activity could account for the unusual behavior of the monetary aggregates or electricity consumption, distorting the estimates yielded for NOA activity.

  • Latent variable based estimates of NOA activity tries to address the shortfalls in the approach above. They attempt to isolate the “grey economy” factors from others in accounting for unusual observations in velocity or other indicators. Thus “grey economy encouraging factors”—such as tax and regulatory burdens and citizens’ attitude to tax compliance—are included as independent variables in regressions explaining money demand or electricity consumption. The estimates of NOA activity are derived from the parameter estimates for “grey economy inducing variables” in such equations. Lackó (2000) applies this method to electricity consumption.

6. Internationally, indirect-based estimates of NOA are typically larger than those produced by official direct (survey) based methods. This is evident from the fact that even when official published GDP data include “direct” adjustments for NOA, the indirect estimates call for yet further adjustments. The additional adjustments suggested are typically lowest in the developed countries, higher in middle income countries, higher still in the CIS area and highest in the Former Yugoslav area (Text Table 2). It is an open—and much disputed—question, however, if the “indirect” estimates are reliable.

Text Table 2.

NOA Estimates in Transition Economies

(In percent of GDP; unless otherwise specified)

article image
Source: Schneider (2005), LackÓ (2000)

Excluding Bosnia and Herzegovina

Belgium, Canada, Ireland, Italy, and USA. LackÓ estimates are for 1990.

C. NOA in Bosnia and Herzegovina

7. BiH makes no adjustment for NOA in the official published GDP numbers. Lack of administrative resources and proper organizational incentives has delayed initiatives to address this shortfall. In this context, some foreign donors have produced their own “informal” estimates of NOA activity in BiH, but these have not been reflected in any official agencies’ data.

8. BiH’s NOA implied by non-official application of “direct methods” appears to be similar to those implied by official adjustments to data in CIS economies. The estimates of the IMF Statistical Advisor in BiH suggest that NOA was around 40 percent of unadjusted (official) GDP in 1999-2002, albeit on a declining trend.4 This would put the NOA at some 28 percent of total—officially published GDP plus NOA—activity, suggesting that NOA in Bosnia and Herzegovina is proportionately close to the CIS group and above those for other transition and former Yugoslav countries (Table 2).

9. A more recent study, applying the “direct method” to BiH assessed NOA at 58 percent of official GDP in 2001 (Text Figure 1).5 This would put the NOA at some 40 percent of total activity, much higher even than the official adjustments using direct methods that are made in the CIS area. This study also attempts a decomposition of the NOA sector (See annex 1). This finds that only around seven percentage points of GDP of their NOA estimate concerns illegal activities. This suggests that shortcomings in the statistical system, rather than illegal activity, overwhelmingly account for difficulties in reflecting NOA in official data (Text Figure 1). On these estimates, the largest share of NOA comprises the shadow economy. Enterprises underreporting revenue, companies using false registration numbers, and small merchants produce a third of BiH’s and more than 40 percent of Republika Srpska’s (RS) officially published GDP.

Text Figure 1.
Text Figure 1.

Bosnia and Herzegovina: Structure of NOA in 2001

(In percent of respective officially published GDP)

Citation: IMF Staff Country Reports 2005, 198; 10.5089/9781451977783.002.A005

Source: Dell’Anno and Piirisild (2004)

10. However, while indirect methods applied internationally generally imply that official adjustments for NOA are too cautious, this does not appear to be the case for BiH (See Text Table 2). Two sets of NOA estimates applying the latent variable approach to BiH are available.6 According to both of them, the grey economy was 30–37 percent of officially published GDP in 1999–2003, lower than—rather than well above as is typical in international experience—the available estimates yielded by “direct” methods. The reasons for this are not clear. Nevertheless, note also that the two indirect estimates differ regarding the trends, with Schneider (2005) setting it on an increasing trend, while Dell’Anno and Piirisild’s (2004) estimates suggest that it has decreased slightly since 2002 (Text Figure 2).

Text Figure 2.
Text Figure 2.

Bosnia and Herzegovina: Estimates of NOA, 1999-2003

(In percent of officially published GDP)

Citation: IMF Staff Country Reports 2005, 198; 10.5089/9781451977783.002.A005

Source: Dell’Anno and Piirisild (2004)

D. New Estimates of NOA

11. Both direct and indirect approaches suggest significant NOA in Bosnia and Herzegovina. But, surprisingly, the direct estimates are higher than those yielded by indirect methods, raising some doubts as to the reliability of both. In this light, a further approach was attempted to estimate NOA in BiH.7

12. Extending the one-variable methods using currency in circulation and electricity consumption, it uses international panel data to estimate the “typical” correlation between such variables and GDP. The values of those variables for Bosnia and Herzegovina are then inserted into the equation, yielding the “estimated GDP for BiH.” The difference between that and “official GDP” is the implicit estimate of NOA.

13. The exercise proceeded in several steps. Variables for which observations are available after 1999 in BiH were identified—such as telephone lines, health, and macroeconomic indicators. Data on the same variables for a panel of some 193 countries were assembled, and the bivariate correlations of each with nominal GDP were calculated and those with low bivariate correlations were discarded. The variables remaining were included in a large regression, and progressively eliminated using a standard “step down” procedure until a parsimonious regression remained (Text Table 3).

Text Table 3.

Pooled Regression with Robust Standard Errors, 1999-2002

article image

14. Only a few variables have high explanatory power. Among them health care expenditures, exports, and currency in circulation “account for” more than 90 percent of the variation in GDP.8 The set of “significant bilateral correlation” variables was available for 96 countries in the period 1999–2002, among them 22 transition economies, 44 low-middle income countries, and 22 conflict and post-conflict countries.9

15. On this basis, NOA in BiH is significant. Between 1999 and 2002, the available indicators fitted to the full sample of available countries suggest that official GDP estimates in BiH should be increased by 30–33 percent (Text Table 4, Text Figure 3a). This is the level of activity which would be expected from that array of indicators based on international experience.

Text Figure 3a.
Text Figure 3a.

Bosnia and Herzegovina: Size of NOA Based on 96 Countries

(In percent of official GDP)

Citation: IMF Staff Country Reports 2005, 198; 10.5089/9781451977783.002.A005

Text Table 4.

NOA in Transition Economies Relative to the Rest of the World, 1999-2002

(In percent of official GDP)

article image
Source: IMF Staff Estimates

16. Note this is not to say that NOA is some 1/3 of official BiH GDP. Two considerations apply here.

  • First, the exercise reports that those health and other indicators in the regression in BiH would, internationally, be associated with “official” GDP estimates 1/3 higher than is the case in BiH. Given that evidence from indirect estimates is that internationally, GDP is also underestimated, the implied NOA could be a higher proportion of BiH published GDP data still.

  • Second—and this goes in the opposite direction—the implied discrepancy between published and “estimated” GDP could reflect factors other than NOA. For example, if some characteristic of the BiH economy compromises efficiency by more than is typical in international experience—such as post-conflict complications or X-inefficiencies in the corporate sector—then true GDP in BiH will be below levels which would otherwise be predicted on the basis of international experience, given the BiH scores for health and other indicators.10 The intuition here is that “potential GDP”, as suggested by these health and other indicators, could be above “actual” BiH GDP. If so, then the discrepancy between estimated and published BiH GDP exaggerates NOA.

17. With these key caveats, further conclusions can be drawn. The regression was reestimated excluding all transition economies, and it was used to “predict” BiH GDP. On this basis, official GDP in BiH would be expected to be some 45–50 percent higher than reported, based on non-transition international experience (Text Table 4, column 2). This, along with the first result, implies that underreporting (or output impairing inefficiencies) is systematically heavy in the transition area. This possibility is also suggested by an exercise re-estimating the regression using non-transition countries, and comparing “forecast” with “published” GDP for all transition countries’ GDP (Text Table 4, column 3). And a similar exercise confirms for Former Yugoslav countries suggests that they also significantly underreport activity compared with international non-transition experience.

18. Moreover, within the cluster of transition economies, BiH’s NOA has increased in 1999–2002. This finding suggests that BiH lags behind other transition countries in meeting OECD’s exhaustiveness criteria for GDP compilation, in combating grey activity, or/and in curtailing production inefficiencies (Text Figure 3b). Similarly, BiH exhibits high NOA estimates relative to low middle income and conflict and post-conflict economies, respectively (Text Figures 3c–3d).

Text Figure 3b.
Text Figure 3b.

Bosnia and Herzegovina: Size of NOA Based on 22 Transition Economies

(In percent of official GDP)

Citation: IMF Staff Country Reports 2005, 198; 10.5089/9781451977783.002.A005

Text Figure 3c.
Text Figure 3c.

Bosnia and Herzegovina: Size of NOA Based on 44 Low Middle Income Countries

(In percent of official GDP)

Citation: IMF Staff Country Reports 2005, 198; 10.5089/9781451977783.002.A005

Text Figure 3d.
Text Figure 3d.

Bosnia and Herzegovina: Size of NOA Based on 22 Post-Conflict Countries

(In percent of official GDP)

Citation: IMF Staff Country Reports 2005, 198; 10.5089/9781451977783.002.A005

Source: IMF Staff estimatesNote: Grey lines represent two standard deviations around NOA estimates.

E. Concluding Observations

19. Considerable evidence points to understatement of activity in official BiH national accounts data. Direct, indirect, and the staff methods—just described—all point in this direction. Staff estimates—30–50 percent—straddle the IMF Statistical advisor’s and other “direct” estimates prepared for BiH, reported in ¶8–9. However, this “confirmation” of the range of those estimates, even in the face of direct and indirect estimates of similar magnitude, is qualified to the extent that our methodology does not allow estimates for NOA to be distinguished from deviations of actual from potential output.

20. On this basis, the case to commence efforts in BiH to make adjustments to official data for NOA are strong. Even if, like countries in the region, these adjustments are modest initially, this would strengthen the national accounts statistics. Efforts to set up surveys etc which could form the basis for such “direct method” adjustments would therefore be appropriate.

21. But the evidence that NOA is disproportionately large in BiH is mixed. The direct and indirect measures, as well as the staff estimates, put BiH somewhere in the middle of the CIS group, although this group does seem to have disproportionately large non-observed activity compared with other countries internationally. And while the staff approach suggests that NOA in BiH is high relative to transition and Former Yugoslav cases, this conclusion is qualified by the “potential output issue.”

22. Subject to this last qualification, the available evidence implies that economic indicators for BiH measured as ratios to GDP, including current account deficit and government spending, are likely to be overstated. But in assessing the implications for interpretation of such indicators, caution is warranted. Those other data, like the GDP data, also have shortcomings.

  • So while an “upward revision” to GDP estimates would lower the estimated external current account deficit relative to GDP, it should be recalled that merchandise imports are also likely significantly underreported. For example, using the Central Bank measure of the current account balance in euros—which adjusts for this factor—and expressing that relative to “adjusted” GDP 50 percent higher than the official estimate, yields a deficit of 16 percent of GDP, close to the current staff estimate of 17 percent of GDP.

  • Similarly, though on staff data—used because the authorities do not compile data on this key aggregate—consolidated government spending accounts for some 50 percent of GDP, this takes no account of road funds and other extra budgetary funds below entity central government level, nor of quasi-fiscal activities which could be considerable, notably in the corporate sector.11

23. All these considerations underscore the need to strengthen statistics to provide better guidance for policymaking and surveillance. This would form a key par of BiH’s strengthened fiscal architecture.12

Annex 1. Decomposing NOA

Dell’Anno and Piirisild (2004) estimated, for 2001, all NOA elements in BiH except the effect of outdated registries and omission of businesses for statistical reasons.

  • To estimate the size of NOA due to non-response, Dell’Anno and Piirisild used information from the Federation Bureau of Statistics, which evaluated that non-response was around 10 percent of surveyed units. The same percentage of non-response was assumed to hold in the RS. Non-response accounted for 5.9 % of official GDP.

  • Underreporting of value added was calculated using tax inspections data from the RS Tax Administration. Undeclared sales amounted to seven percent of large tax payers’ declared sales, or, as roughly estimated, 21 percent of declared sales of all tax payers. Underreporting, adjusted with an estimate for restaurant tips, was evaluated at 12.7 percent of official GDP.

  • Using Federation tax audit data, unregistered activities were estimated as the ratio of discovered underground businesses to active legal entities. The underground value added amounted to 2.3 percent of official GDP.

  • The size of the informal sector was estimated by reducing labor productivity in the formal sector and multiplying it by the number of informal workers from World Bank (2002) data. The procedure was carried out for each branch of the economy and each entity separately. With an additional assessment of actual and imputed rents, the size of the informal sector was estimated at 29.4 percent.

  • Illegal production estimate was based on information about the average bribe amount and the percentage of the population having paid bribes, as well as a CBBH’s assessment of the value added of prostitution activities. Illegal production was evaluated at 7.3 percent of official GDP.

References

  • Beckerman, Wilfred, and Robert Bacon, 1966, “International Comparisons of Income Levels: A Suggested New Measure,The Economic Journal, Vol. 76, No. 303, pp. 51936.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dell’ Anno, Roberto, and Marje Piirisild, 2004, “Estimate of Non-Observed Economy in Bosnia and Herzegovina.” (unpublished: paper presented in Sarajevo, Bosnia and Herzegovina, April)

    • Search Google Scholar
    • Export Citation
  • Organization for Economic Cooperation and Development, 2002, Measuring the Non-Observed Economy: A Handbook (Paris).

  • United Nations, 2003, Non-Observed Economy in National Accounts: Survey of National Practices (Geneva).

  • Fleming, Mathhew H., John Roman, and Graham Farrell, 2000, “The Shadow Economy,Journal of International Affairs, Vol. 53, No. 2, pp 387409.

    • Search Google Scholar
    • Export Citation
  • Gilbert, C.L., Short Cut Methods of Estimating Per Capita GDP and Tentative Results on the Distribution and Growth of World GDP,Oxford Economic Papers, Vol. 34, No. 3, pp. 45272.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kaufmann, Daniel, and Aleksander Kaliberda, 1996, “Integrating the Unofficial Economy into the Dynamics of Post-Socialist Economies: A Framework of Analysis and Evidence,” World Bank Policy Research Working Paper 1691.

    • Search Google Scholar
    • Export Citation
  • Lacko, Maria, 2000, “Hidden Economy—An Unknown Quantity? Comparative Analysis of Hidden Economies in Transition Countries, 1989–95”, Economics of Transition, Vol. 8, No. 1, pp. 117149.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schneider, Friedrich, and Dominik Enste, 2000, “Shadow Economies: Size, Causes, and Consequences,Journal of Economic Literature, Vol. 38, No. 1, pp. 77114.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schneider, Friedrich, 2005, “Shadow Economies of 145 Countries all over the World: Estimation Results over the Period 1999 to 2003,” paper presented at the Annual Meeting of the European Public Choice Society, Durham, UK, April.

    • Search Google Scholar
    • Export Citation
1

Prepared by Iva Petrova

2

IMF Country Report No. 04/54, March 9, 2004.

3

For a detailed survey of the methods see Fleming, Roman, and Farrell (2000), Lackó (2000), OECD (2002), Schneider and Enste (2000), and Schneider (2005)

4

IMF Country Report No. 04/54, March 9, 2004.

5

Dell’Anno and Piirisild (2004) for the Financial Services Volunteer Corps.

7

The approach builds on Beckerman and Bacon (1966) and Gilbert (1982).

8

Note that causality cannot be inferred from this regression—good health does not “cause” high GDP—because no corrections for endogeneity have been made. The exercise yields correlations, not causation.

9

According to the World Bank’s income classification, BiH is in the group of low-middle income countries. The armed conflict classification is from the Center for the Study of Civil War: http://www.prio.no/cwp/armedconflict.

10

See Chapter 4.

11

See Chapter 4.

12

See Chapter 10.

Bosnia and Herzegovina: Selected Economic Issues
Author: International Monetary Fund
  • View in gallery

    Bosnia and Herzegovina: Structure of NOA in 2001

    (In percent of respective officially published GDP)

  • View in gallery

    Bosnia and Herzegovina: Estimates of NOA, 1999-2003

    (In percent of officially published GDP)

  • View in gallery

    Bosnia and Herzegovina: Size of NOA Based on 96 Countries

    (In percent of official GDP)

  • View in gallery

    Bosnia and Herzegovina: Size of NOA Based on 22 Transition Economies

    (In percent of official GDP)

  • View in gallery

    Bosnia and Herzegovina: Size of NOA Based on 44 Low Middle Income Countries

    (In percent of official GDP)

  • View in gallery

    Bosnia and Herzegovina: Size of NOA Based on 22 Post-Conflict Countries

    (In percent of official GDP)