>>>Back to Home
 Opening Remarks
Panel 1: How Do We Measure Productivity?
Panel 2: Is There a Job Growth – Productivity Tradeoff?
Panel 3: Firm and Farm-Level Dynamics and Productivity.
Panel 4: What Are the Sectoral Drivers of Productivity Growth?
Closing Remarks
cvcxzvzxzvcxvxczvcxzvxzcvzcxvzxvxzcvxczvzxvxzcvxvxczvx cnmcnvcvmnvbnmb vbnmmnmnzv
Opening Remarks
Ms. Cheryl W. Gray (Sector Director, ECSPE) welcomed presenters, commentators, and other participants. This workshop was meant to catalyze Bank thinking on an 18-month research program on productivity and growth in ECA. She dwelled upon the two-pronged motivation for looking at productivity issues. First, the economic transition is about the shift of resources from low productivity to higher productivity uses, and it is therefore important to examine current trends in productivity and to suggest areas for future reform. Second, the recently completed regional study on poverty showed that real wage increases have been an important driver for poverty reduction over 1998-03. But this driver can only be sustainable as a source for poverty reduction if productivity also increases at the same rate. She encouraged the workshop participants to participate fully in the discussions and to brainstorm on the feasible research agenda for the proposed regional study.
Panel 1: How Do We Measure Productivity?
Productivity indicators and the data sources they come from should correspond to the specific research questions being asked. For the analysis of monetary policy problems, researchers need indicators of potential output (labor, capital services, TFP), resource bottlenecks (unit labor costs), and business cycle (cyclicality of productivity, prices and wages). On the other hand, examination of structural policy problems requires performance indicators, such as productivity comparisons across sectors or countries and over time as well as program evaluation techniques.
There are numerous problems with both the input and the output sides of productivity indicators, and the measurement errors in one do not cancel out the measurement errors in the other. Micro-level and macro-level data sources are often significantly inconsistent and not integrated. Labor data used in productivity indicators come from different types of surveys and often do not add up to the same aggregate. Lack of separate deflators for inputs and outputs make time series data difficult to create, while lack of PPP make cross-country comparisons unreliable. Keeping these problems in mind, it is essential to test robustness of findings across different productivity measures and understand which dimensions of productivity matter for a given result to hold.
A study of indirect costs of production (transport, telecommunications, water, security, bribes) in African countries was presented to illustrate application of productivity and business climate data from Investment Climate Assessment surveys. The main implication of this study is that a focus on reducing indirect costs would have a very significant effect on profits, since even firms with high gross TFP suffer from reduced employment potential and lower value added due to indirect costs. Market segmentation may also lead to lower productivity.
Allocative efficiency, which can be measured as a static concept (how well resources are allocated, estimated by Olley-Pakes decomposition) or a dynamic concept (either due to changes over time or due to entry and exit of firms), accounts for much of the observed cross-country dispersion in productivity. To obtain an accurate estimate of allocative efficiency, one needs to measure productivity (both macro and micro) consistently as well as to track firm dynamics (entry and exit), both of which are challenging to do. Since labor productivity and TFP are often highly correlated, it might be more important to get a representative sample of enterprises and firm dynamics than to estimate TFP accurately.
World Bank’s ICA surveys are a source for productivity data that also includes business climate variables but their applicability for estimating allocative efficiency is uncertain for now. The representativeness of the sample differs between the two types of ICA surveys: the first covers the whole economy with smaller samples in each sector and fewer questions on productivity while the second covers only key manufacturing sectors but with deeper analysis of productivity. The ideal survey would have both greater representation of sectoral coverage as well as deeper analysis of productivity. The latest BEEPS of the ECA region conducted together with EBRD has improved representativeness. ICA surveys have a floor for firm size and they oversample large firms, but this could be corrected with appropriate sample weights. As for firm dynamics, ICA surveys are only entering the second round, but this round will look at exit trends and link up with census data; thus, this aspect of the surveys will improve with time. The ECA region, which has very good access to official statistics, pioneered the work on BEEPS surveys and has already completed the third round, which includes a panel component.
While using existing surveys, we should work with country statistical agencies to build a statistical roadmap to improve productivity measurement. To improve the availability of data on productivity, the EU should continue collecting data for the sectoral productivity database (EU KLEMS), while other countries should adopt a similar approach but obtain data from original sources and harmonize different sources and methods to create consistent integrated data. Focusing on sectoral and firm-level data will provide answers to interesting research questions, such as R&D, ICT, corporate taxes, and many other topics. We need to improve the sampling of entry and exit of firms as well as organizational capital. And we need to recognize cross-country diversity in the structure of the private sector (Bulgaria’s large companies versus Romania’s SMEs), and analyze the productivity not only of the private but also of the public sector.
There may be different ways of looking at sectoral issues. One could look at classical sectors—agriculture, industry, and services—and subsectors within them. Or one could define the sectors in terms of their level of capital intensity or formal and informal or urban and rural or market and non-market. It is also important to compare countries at the same stage of restructuring.
In order to understand productivity, we need to (i) define the policy questions to address; (ii) approach the issues through a “triangulation” of macro, sectoral, and micro level evidence; (iii) recognize the importance of indirect costs in reducing productivity; (iv) seek panel data, so as to take into account entry and exit; (v) work with countries to develop a statistical road map as data process may be as important as the analysis; and (vi) focus on trends in allocative efficiency, especially on whether forms are moving into higher productivity uses.
Panel 2: Is There a Job Growth – Productivity Tradeoff?
Regardless of whether there is a job growth–productivity tradeoff in the short- and medium-term, it is possible to achieve a long-run equilibrium that would have both greater employment and productivity. Short-run and medium-run tradeoff between job growth and productivity can result from decreasing marginal returns to labor, adjustment to business cycle fluctuations, and skill heterogeneity of workers – the unemployed are those with lower skills, so once the unemployed start working, average productivity per worker falls. The tradeoff could also arise due to firm restructuring or structural shifts in the economy, as production moves from labor-intensive activities with lower productivity (e.g. manufacturing) to capital-intensive activities with higher productivity (e.g. ICT). Structural shifts in ECA do not follow the same pattern: as Czech Republic’s production moved to services from agriculture and industry, Kyrgyz Republic witnessed an expansion of the agricultural sector. The tradeoff is not necessarily observable at the aggregate level – there might be inter-sectoral employment shifts due to technological change, which leave macroeconomic employment indicators unaffected. Productivity growth in some sectors (like manufacturing) is more likely to result in the tradeoff than in other sectors (transport and communication). Some evidence for a persistent tradeoff between employment and productivity does exist. The negative correlation between hours worked and productivity can be observed in the data comparing the US and the EU, as well as the country-level data within the EU and industry-level data in Italy. However, some EU countries, like Ireland and Finland, were able to achieve high growth rates in both employment and productivity, while other countries, like Germany and Switzerland, exhibited slow growth in both dimensions. The countries in the ECA region would benefit from studying success cases, like that of Ireland. It would also be useful to study clusters of countries to see the common characteristics that drive these relationships.
Since productivity growth can be the effect of restructuring, a particularly high negative correlation could be a sign of fast technical change and/or mismatch of old abilities and new demands, thus it is not necessarily a negative phenomenon. Indeed, structural reforms and improvements in allocative efficiency are essential to ensure that in the long run, both employment and productivity are higher. While structural reforms can decrease employment in the short run, free entry into the product market lowers rents, increasing the number of firms with higher output and employment. Entry and exit of firms play a particularly significant role in transition economies, since in these countries new firms are more productive than existing firms. It would be useful to examine whether structural shifts in the economy have led to convergence in production and wages across sectors.
To provide “decent work” before economies achieve the favorable long-run equilibrium, policymakers have to ensure that dislocated workers are adequately compensated and that there are backward and forward linkages between dynamic sectors. Creating an unemployment benefit scheme at the beginning of the transition period was vital for the productivity growth in CEE countries. The absence of such benefit made CIS economies more resistant to undertaking productivity enhancements that result in labor shedding and in displaced workers moving to subsistence agriculture instead of higher-productivity activities. While employment schemes may be appropriate instruments to provide safety net for displaced workers in some countries, cash transfer initiatives might be more fiscally prudent in other countries, such as those in ECA.
Efforts should be targeted to formalize employment in the region, as informal work has lower productivity and is not covered by the labor market protection policies. Finally, attention should be paid not only to employment growth, but also to wage growth. For example, ECA’s recent improvements in productivity translated into high wage growth but not employment growth. It is important that wage growth does not step out of sync with productivity growth.
Several conclusions were drawn by the Chair. These included:
(i) take full account of creative destruction;
(ii) look at intersectoral linkages;
(iii) examine the role of market segmentation and informality;
(iv) look at the role of structural policies as they impact upon product and factor markets separately;
(v) look at his outliers—successful on both jobs and productivity front; and
(vi) examine the role of education in fostering a win-win situation.
Panel 3: Firm and Farm-Level Dynamics and Productivity.
Firm-level (and farm-level) productivity analysis provides information on the effect of firm and farm size, firm and farm ownership structure as well as the effect of different policies and macroeconomic variables on firm-level productivity. Firm-level analysis can be conducted using structural models as well as reduced-form models or productivity decomposition. While the World Bank has focused on studies of aggregate productivity level and growth rate, the studies presented in this panel point to the importance of firm-specific productivity drivers. Besides, ECA has already significantly improved its macroeconomic business environment, so the marginal returns to further advances on this front may be decreasing. Thus, we should focus on country-specific firm-level and sector-level work.
The most persistent finding in firm-level productivity studies conducted using data from ECA countries was that foreign-owned firms are more efficient than domestic-owned firms. This result was obtained by two studies using different data sets and different estimation methods – Svejnar and Sabirianova’s study of firm revenues in Czech Republic and Russia using 2SLS and Blundell-Bond GMM estimation as well as a study by Brown, Earle, and Telegdy of the effects of privatization in Romania, Hungary, Ukraine, and Russia using program evaluation (difference-in-difference) methods. Foreign-owned firms were also found to improve efficiency faster, as well as converge to a higher steady state than domestic firms.
Potentially, the above findings could be caused by selection bias (or “creaming) – foreigners could choose to buy firms that are already the most productive. This hypothesis is confirmed by Svejnar and Sabirianova, who found that part of the superior performance of foreign firms is due to selective acquisitions of local firms. Brown et al. also had some selection bias in the sample of privatized firms, but it was not always the case that initially more productive firms were the ones privatized. Thus, studies such as the two mentioned above have to control for selection bias in order to provide more accurate estimates for the effect of foreign ownership and privatization on firm productivity.
On the relationship between structure of ownership and productivity, the results are ambiguous. While private and mixed ownership firms were found to be somewhat more efficient than state-owned enterprises (SOEs) in Czech Republic, they were less efficient than SOEs in Russia. As far as privatization by domestic firms, whereas it had a positive impact on productivity in Hungary and Romania, its effect was ambiguous in Ukraine, and negative in Russia. Hence, we should avoid simple stories about the relationship between firms’ ownership structure and productivity.
Another approach to firm-level productivity analysis is construction of structural forward-looking models of industrial evolution, which incorporate uncertainty and multi-agent optimization. Several of these structural models were presented to illustrate the impact of different variables on turnover-based productivity growth (TBPG), which occurs through entry and exit of firms and through resource reallocation towards more productive firms. The alternative to turnover-based growth is intra-firm productivity growth (IFPG), which happens through vintage effects, endogenous innovation, learning spillovers, and random shocks. The studies using structural models found that subsidies for incumbent firms after a negative macroeconomic shock keep low-productivity firms active and significantly reduce present value of GDP; depreciation discourages investment by raising borrowing costs; during periods of macro volatility households with modest wealth are less likely to operate proprietorships while incumbents of large low-productivity firms are likely to hang on and hope for the best; and reduced import prices initially lead to better quality of domestic goods but then reduce the incentives to innovate due to smaller profits. There is also evidence to suggest that labor elasticities to growth increase with a lag or 2-3 years after growth start. Hence, theoretical models have already begun to provide some answers, and future research should focus on econometric estimation, endogenous innovation, modeling of entrants, accounting for product quality and demand, and entry costs. However, the results from studies using structural models can sometimes be too subtle and too dependent upon time horizon and specific assumptions to be.
A very distinct form of micro-level productivity analysis has been developed for the agricultural sector. Farm-level measurement of productivity uses land and labor productivity as well as TFP. The ECA region has experienced two distinct trends in farm productivity: in most CIS (especially Russia) yields, output, and productivity have been (and are) falling whereas in Central and Eastern Europe all these indicators fell in the beginning of the transition period but then output stabilized and productivity together with yields rose. This disparity can be explained by different intensity of initial shocks and differential implementation of reforms. Analysis of three different patterns of agricultural transition implies that extensive land reforms and hard budget constraints were essential for improving farm productivity, the optimal size of farms depended on technology and endowments, use rights initially provided sufficient incentives, and innovative exchange institutions, availability of off-farm employment as well as sale/lease of land were important for allocative efficiency. In conclusion, this session started unbundling the ‘black box’ to see what matters for firm and farm level productivity. Four points stood out:
(i) policies and institutions matter;
(ii) the many interactions among endogenous variables will challenge the analysis;
(iii) it is important to benchmark ECA countries against successful EU countries; and
(iv) employment elasticities to growth may rise with a lag of 2-3 years.
Panel 4: What Are the Sectoral Drivers of Productivity Growth?
Many drivers of productivity growth are sectoral in nature. The quality of infrastructure, the level of health and education of the labor force, and the role of technology and innovation in the economy, can all contribute to the growth of overall productivity.
Urban locations can be the most promising sources of future sectoral productivity growth as they are able to assemble different stages of production in a tight geographical space. Cities also attract talented and diverse population, which contributes to the development of skill-intensive activities and innovation capability, thereby establishing agglomeration economies, which can already be found in East and South Asia. However, policymakers have to ensure that congestion costs do not undercut the advantages of agglomeration. Thus, sensible city management, efficient land use, and effective service provision are vital for maintaining this productivity driver. Sustaining high-quality university education and building up links between the universities and the private sector will guarantee replenishment of human capital in the cities. ECA as a region is highly urbanized, but many of the cities did not arise from market forces. To develop its industry towns into agglomeration economies, cities in ECA needs to improve investment climate, infrastructure, and management of the urban land market, especially rental housing market, which affects workers’ mobility.
Availability of modern, high-quality infrastructure is essential for productivity growth. Thus, infrastructure reform is considered to be an important step to improving productivity and competitiveness. The impact of the standard recommendations for reforming this sector—creation of an independent regulator and privatization—have been analyzed for electricity, water, and ICT in ECA. The impact of the two reforms on access, affordability, and quality of infrastructure is hard to discern in the region but appears to be positive for energy and transport while remaining ambiguous for water. However, econometric analysis of electricity and telecommunications shows that both reforms tended to raise prices, and these higher prices affected demand, thereby reducing access. Indeed, effective access and quality of infrastructure in ECA is disappearing for some users, especially in secondary cities. Both entry / exit of providers and spatial reallocation affect this trend. And poorer regions seem to lose out. And fiscal costs remain high, e.g. original contracts in LAC were renegotiated with subsidies. Neither of the two reforms—creation of the independent regulator and privatization— had a significant effect on the service quality. However, corruption does appear to decrease after the two reforms are implemented. The implication of this analysis is that improvements in infrastructure depend more on technological progress and other complementary.
Support of innovation and development of the ICT sector can stimulate productivity growth. In CEE countries, the ICT-producing sector is still too small to drive convergence and there is no evidence of spillovers from that sector. It is ICT-using sector that can (and does) contribute substantially to overall productivity growth. Growth of ICT-using services requires flexible product and labor markets, marketable human capital, business reorganization around ICT use and more sophisticated managerial techniques, all of which depend on complicated and socially sensitive reforms. Policymakers in CEE countries should implement deep structural reforms, increase competition, develop e-services and e-procurement, increase public funding in education of ICT use, and promote ICT use through different means.
Innovation can happen both at the knowledge frontier and in diffusion of new technologies; the latter was observed in both Spain and Italy. Innovation cannot be considered in a vacuum as it is closely related to private sector development and accumulation issues. Thus, links between R&D and production have to be strengthened to increase the returns to R&D. Developing countries should focus on increasing the technical intensity of existing production instead of reorienting production to higher-technology goods. One direction for research is explaining TFP as a function of knowledge alongside traditional variables. One could also examine the impact of innovation spillovers and R&D factors on growth but one would have to be concerned with endogeneity.
Productivity growth is difficult to improve without a healthy and educated labor force. However, ECA has witnessed relatively low returns to schooling, which could be explained by underinvestment in human capital and wage caps on skilled labor. Fast-reforming transition economies have seen higher returns to education. In ECA, we need to not only model selection into employment but the choice between public and private sector as well as selection into vocational education (and out of general education), the returns to which are declining. With respect to health, studies have found causal impact of health status on wages and labor force participation rates. Lack of relevant health data for ECA as well as the lack of proxies for health status and instruments to account for endogeneity constrain the research options. Also, government institutions have been shown to have a significant effect on the provision of education and health services.
The Chair concluded by saying that:
(i) it is difficult to be sure on what the drivers of productivity growth are as data on cost and measurement are difficult and subject to various measurement problems and policies vary; (ii) agglomeration economies are likely to be significant with small increases in inputs leading to large returns; and
(iii) there are obvious complementarities between physical and human capital investments.
Closing Remarks
Mr. Pradeep Mitra (Chief Economist, ECAVP) closed the workshop by thanking the participants for a stimulating discussion and making some comments on the upcoming study on productivity in the ECA region. His conclusions included the following:
i. Although the study will be region-wide, we need to take account of diversity within the region by considering five subregions: EU-8, SEE, middle-income CIS, low-income CIS, and Turkey.
ii. There is a huge policy problem. Although the recovery from the 1998 crisis was strong and broad-based, would the drivers of this growth still be available in the future?
iii. The upcoming report needs to lay out the macroeconomic story in the beginning. But we need to consider different sectoral stories of agriculture, industry, and services. However, most of the analysis will be conducted at the firm level.
iv. On data issues, we still need to decide what combination of ICA surveys, BEEPS data and other data sources to use for the report. The report should discuss the important but changing role of net entry.
v. We also need to analyze productivity both from demand and supply perspectives, the latter having to do with access to and quality of public services.
vi. Finally, we should not forget that for low-income CIS countries, agriculture is very important, and we should distinguish between labor-intensive and land-intensive agriculture.
vii. Finally, although we would conduct the analysis for sub-regions, we need to bring in case studies (Russia, Turkey, Poland, low-income CIS).
Mr. Mitra concluded by hoping that the workshop participants would continue to contribute their expertise once the report preparation gets under way.
|
RSS