The impact of the AIDS epidemic on African firms.This paper is a sequel to an earlier study, titled "Impact of the Epidemic on Firms," published in the National Research Council's 1996 book Preventing and Mitigating mit·i·gate
v. mit·i·gat·ed, mit·i·gat·ing, mit·i·gates
To moderate (a quality or condition) in force or intensity; alleviate. See Synonyms at relieve.
To become milder. AIDS in Sub-Saharan Africa. The earlier study used a new data set, provided by the World Bank's Regional Program on Enterprise Development (RPED RPED Retinal Pigment Epithelial Dystrophy ), to examine the impact of AIDS in several severely affected African countries. At the time, data were available from only three of seven African countries included in the World Bank's surveys--Ghana, Kenya and Zimbabwe. Recently, more data have become available which allow us to extend the investigation to several other nations. In addition, the earlier study was restricted to an assessment of enterprise worker attrition Attrition
The reduction in staff and employees in a company through normal means, such as retirement and resignation. This is natural in any business and industry.
Notes: rates associated with HIV HIV (Human Immunodeficiency Virus), either of two closely related retroviruses that invade T-helper lymphocytes and are responsible for AIDS. There are two types of HIV: HIV-1 and HIV-2. HIV-1 is responsible for the vast majority of AIDS in the United States. infection. It did not address the broader issue of the impact of attrition due to HIV on firm performance.
In this study, we add to the results of the earlier National Research Council investigation in two ways. First, we expand the examination of worker attrition associated with HIV infection to two of the most seriously affected countries in Africa, Zambia and Tanzania. Second, we use observed worker attrition rates due to HIV infection to investigate the effect of AIDS on firm performance. Following a brief introduction, we examine the effect of the HIV/AIDS epidemic on enterprise worker attrition in five countries in East and West Africa West Africa
A region of western Africa between the Sahara Desert and the Gulf of Guinea. It was largely controlled by colonial powers until the 20th century.
West African adj. & n. . Next, we investigate the impact of AIDS on several measures of firm performance, including enterprise productivity and profitability.
The data we use to examine these issues comes from a random sample of approximately 200 manufacturing firms in each country, operating in four sectors (food processing Food processing is the set of methods and techniques used to transform raw ingredients into food for consumption by humans or animals. The food processing industry utilises these processes. , metal working, wood working and textiles & garments). The survey of these firms was a longitudinal lon·gi·tu·di·nal
Running in the direction of the long axis of the body or any of its parts. exercise, conducted over a three-year period beginning in 1992. The aim of the research program was to investigate the dynamics of enterprise manufacturing growth in Africa, especially with respect to firm response to government policy reforms supported by the World Bank and other international agencies. At the request of researchers interested in investigating the impact of AIDS, several questions were added to the survey in years two and three, designed explicitly to learn about worker attrition rates due to sickness SICKNESS. By sickness is understood any affection of the body which deprives it temporarily of the power to fulfill its usual functions.
2. Sickness is either such as affects the body generally, or only some parts of it. or death, to categorize cat·e·go·rize
tr.v. cat·e·go·rized, cat·e·go·riz·ing, cat·e·go·riz·es
To put into a category or categories; classify.
cat the skill levels of the workers who left, and to measure the impact of this attrition on firm performance.
Assessing AIDS' Impact
It is clear that the definitive measure of the impact of the AIDS epidemic is its effect on the well-being of households and individuals, not firms. However, because firms provide employment and organize the production and distribution of goods and services In economics, economic output is divided into physical goods and intangible services. Consumption of goods and services is assumed to produce utility (unless the "good" is a "bad"). It is often used when referring to a Goods and Services Tax. in the economy their performance has an important influence on the welfare of households and individuals. Therefore, it follows that firms are relevant to assessing AIDS' ultimate impact, particularly in terms of the economic costs of the epidemic.
One would expect that the most direct measurable effect of HIV/AIDS on firm performance would come from worker attrition rates due to sickness or death. Additional effects could result from lower worker productivity, as sick workers may be less productive workers. But, empirically, the worker productivity effect is difficult to isolate isolate /iso·late/ (i´sah-lat)
1. to separate from others.
2. a group of individuals prevented by geographic, genetic, ecologic, social, or artificial barriers from interbreeding with others of their kind. and to measure. Also, the AIDS epidemic may have demand side effects Side effects
Effects of a proposed project on other parts of the firm. resulting from declining numbers of consumers for the firm's products or from shifts in expenditures away from particular products and towards health services health services Managed care The benefits covered under a health contract . But demand effects too are difficult to evaluate. Hence, no attempt will be made in the present analysis to assess productivity or demand effects.
Labor attrition influences firm performance because it raises costs. As greater numbers of workers leave, the firm must expend ex·pend
tr.v. ex·pend·ed, ex·pend·ing, ex·pends
1. To lay out; spend: expending tax revenues on government operations. See Synonyms at spend.
2. more resources searching for and training new workers. In addition, if high-level managers are among those leaving, strategic management of the firm might be affected. This could have dire consequences for the firm's competitiveness and perhaps ultimate survival. Therefore, for attrition rates caused by AIDS to adversely affect firm performance, two propositions must be verified, namely: (a) that attrition resulting from illness and death due to HIV infection must be shown to be a large proportion of the total worker attrition in firms; and (b) that the workers leaving due to sickness or death must be shown to be costly to replace or to have a strategic influence on firm performance. On this second point, even if attrition rates were found to be high due to AIDS, recruiting replacements in Africa may be relatively inexpensive because of the existence of unemployed labor. It is often suggested that hiring a replacement worker with equivalent skills in Africa is as easy as taking the first person in line at the front gate of the factory. This may be especially true today as many countries are involved in policy reform programs which are causing large economic adjustments and temporary increases in excess capacity and unemployment.
We assess in the remainder of this paper the extent to which both propositions are applicable in the context of African manufacturing. The case for a significant impact of AIDS on African firms rests on the magnitude of worker attrition and the cost of this attrition to firms.
Worker Attrition Due to HIV Infection
Table 1 presents data on worker attrition in manufacturing firms in five African countries, together with urban seroprevalence seroprevalence Immunology The proportion of a population that is seropositive–ie, has been exposed to a particular pathogen or immunogen; the seropositivity of a population is calculated as the number of individuals who produce a particular antibody divided estimates for HIV infection. Table 2 presents these same data classified by firm size. The figures in Tables 1 and 2 are for 1994 except for the overall rates of attrition which are for both 1993 and 1994. (1)
A total of 992 firms employing 115,136 workers are represented in the sample. Total worker attrition, as a percentage of the work force, shows considerable variation across countries, from about 8% in Kenya to almost 21% in Zambia. Further, as indicated by the 1993 figures in parentheses, overall attrition rates can also vary across time within the same countries. The reason is that total worker attrition varies for many reasons other than AIDS, including the economic conditions prevailing at the time of the survey, as well as factors such as the size distribution of firms in each country. The influence of firm size is illustrated by a comparison of countries like Ghana with Zimbabwe. Average firm size in Zimbabwe, as measured by the number of workers employed, is 300 and in Ghana 50. Because smaller enterprises tend to have higher labor turnover rates in all countries, part of the reason why Ghana's total attrition rate Noun 1. attrition rate - the rate of shrinkage in size or number
rate of attrition
rate - a magnitude or frequency relative to a time unit; "they traveled at a rate of 55 miles per hour"; "the rate of change was faster than expected"
is higher than Zimbabwe's in 1994, notwithstanding its lower HIV prevalence rate, is due to the firm size effect on attrition. This is evident in the first column of Table 2. Changing economic conditions explains a large part of the variation in overall attrition rates in Tanzania between 1993 and 1994. The large figures for 1994 are related to the shrinking of larger firms, many of which were parastatals undergoing privatization privatization: see nationalization.
Transfer of government services or assets to the private sector. State-owned assets may be sold to private owners, or statutory restrictions on competition between privately and publicly owned as part of the structural adjustment program adopted in 1990.
Understanding the effect of AIDS in each country begins with an assessment of firm data on the percentage of attrition due to worker sickness or death. These attrition rates are shown in the last column of Table 1. To evaluate the influence of HIV on these rates, we must compare them with what we expect to observe given the severity of the AIDS epidemic in each country.
Under the assumption that the HIV infection rates for low-risk urban populations, indicated in column one of the table, apply to the manufacturing worker populations in each country, one can estimate the likely observed AIDS-related death, sickness and attrition rates in the work force. (2) For example, using the National Research Council's formula, if the HIV seroprevalence rate were 24.7 percent, as in Zambia, this would mean that 2.5 percent of workers would be forced to leave the work force each year because of AIDS. In addition, for each of these countries it is estimated that 0.5 percent of workers would be forced to leave the work force each year for other health-related reasons. Therefore, in Zambia, a total of 3 percent of the work force would be expected to exit each year due to sickness and death. In Zimbabwe, given its seroprevalence rate of 20.5 percent, attrition due to all health-related reasons is expected to be 2.6 percent, Kenya 2.2 percent, Tanzania 2.1 percent and Ghana 0.7 percent. Comparing these estimates with the attrition rates due to sickness and death found in our sample, one observes that the rank correlation In statistics, rank correlation is the study of relationships between different rankings on the same set of items. It deals with measuring correspondence between two rankings, and assessing the significance of this correspondence. across the five countries is high: the countries with the highest seroprevalence rates seroprevalence rates (sir´ōprev´-lns),
n. also have the highest attrition rates due to sickness and death. Notwithstanding this positive correlation Noun 1. positive correlation - a correlation in which large values of one variable are associated with large values of the other and small with small; the correlation coefficient is between 0 and +1
direct correlation , however, the attrition rates due to sickness and death are found to be systematically lower than would be expected.3 Just as the earlier three-country National Research Council study found, our data suggest that there is a statistically measurable increment To add a number to another number. Incrementing a counter means adding 1 to its current value. to the worker attrition rate associated with HIV infection, but the magnitude of the effect continues to be extremely small, perhaps less than half what would be predicted based on national seroprevalence.
There could be several reasons why the magnitude of the attrition effect is low, while seroprevalence is high. First, the effect of HIV infection may grow as more infected in·fect
tr.v. in·fect·ed, in·fect·ing, in·fects
1. To contaminate with a pathogenic microorganism or agent.
2. To communicate a pathogen or disease to.
3. To invade and produce infection in. workers develop full-blown AIDS. Second, the data on worker attrition due to sickness and death may be an underestimate of the true attrition, because workers who departed for other stated reasons really left due to AIDS. Third, labor laws in most African countries cause firms to avoid "permanent" workers and to hire mostly "casual" workers. Casual workers are flexible in that they enter and depart the firm's work force frequently and very few records are kept on their attendance. Generally, casual workers are paid in cash with no benefits. One effect of all this casual labor is that attrition records are not very accurate; basically, only the permanent employees are included.
Costs of Worker Attrition Due to Sickness and Death
Even if observed attrition rates due to sickness or death are underestimates of the true attrition resulting from AIDS-related causes, the extent to which this is significant depends upon the second factor mentioned earlier; namely, whether or not higher rates of work force attrition have a significant adverse impact on firms' economic performance.
To investigate the cost of worker attrition, we first examine data on the types of workers departing de·part
v. de·part·ed, de·part·ing, de·parts
1. To go away; leave.
2. To die.
3. , together with information on the resources expended ex·pend
tr.v. ex·pend·ed, ex·pend·ing, ex·pends
1. To lay out; spend: expending tax revenues on government operations. See Synonyms at spend.
2. to replace them. Our survey data contain information on a sample of the workers who left the firm (the first 9 workers who departed during the year were chosen in each firm and their personnel records were examined). Of the 1,325 workers who departed, information is available on 439 individuals regarding their skill categories and results of the replacement process. Table 3 reports this information.
Among the workers leaving firms because of sickness and death, 3.6 percent were categorized cat·e·go·rize
tr.v. cat·e·go·rized, cat·e·go·riz·ing, cat·e·go·riz·es
To put into a category or categories; classify.
cat as professionals, 20.3 percent skilled and the remaining 76 percent as semi-skilled or unskilled. This roughly coincides with the skill structure of the work force in most firms. Thus, the limited evidence we present here does not support the popular hypothesis that AIDS may be more heavily impacting richer, more educated segments of the work force. Moreover, as might be expected a priori a priori
In epistemology, knowledge that is independent of all particular experiences, as opposed to a posteriori (or empirical) knowledge, which derives from experience. , the figures in Table 3 also indicate, on average, that it appears to be easier to replace less skilled workers. Only 25 percent of professional workers had been replaced by the time of the interviews, compared with about 47 percent in the other, less skilled worker categories. Also, the proportion of openings that the employer was still trying to fill declined from 37 percent of professionals and 13 percent of skilled workers to 4 percent for unskilled workers. Note that the response is quite high to "decided not to replace" for all categories, reflecting the current downturn in economic activity in these countries. The last column of Table 3 also points to the labor market labor market A place where labor is exchanged for wages; an LM is defined by geography, education and technical expertise, occupation, licensure or certification requirements, and job experience being tighter for higher skilled workers. It took an average of 23 weeks to find a replacement for a professional, 3 weeks for skilled and semi-skilled workers and only 1.9 weeks for unskilled workers. Hence, the hypothesis that search costs Search costs
Costs associated with locating a counterparty to a trade, including explicit costs (such as advertising) and implicit costs (such as the value of time). Related: Information costs. are higher in Africa as the skill category increases receives support here.
The costs of work force attrition are not confined con·fine
v. con·fined, con·fin·ing, con·fines
1. To keep within bounds; restrict: Please confine your remarks to the issues at hand. See Synonyms at limit. to search costs for replacement, however. Firms often undertake worker training, creating firm-specific human capital, and learning-by-doing over the years also raises firm-specific human capital as worker experience on the job increases. Furthermore, depending upon the complexity of the production activity, the arrival of a new person can require other employees to reallocate Verb 1. reallocate - allocate, distribute, or apportion anew; "Congressional seats are reapportioned on the basis of census data"
allocate, apportion - distribute according to a plan or set apart for a special purpose; "I am allocating a loaf of time away from production to training of the new arrival. How substantial are such attrition-related costs in African manufacturing? Table 4 below shows summary measures of age at death and of number of years with the firm--a proxy for firm-specific human capital--for workers who departed firms because of death (about 90 percent of the attrition due to sickness or death).
The average age of unskilled workers and operators at the time of death is shown to be about 40 years, while that for skilled workers is 36. The mean age at death of the 10 professionals in the sample is higher, at almost 47.5. Average experience of all these workers with the firms is close to ten years, which, in terms of acquisition of firm-specific human capital, is substantial. Depending upon the complexity of production activities and of the organization in the firm, loss of skilled workers with such long firm-specific experience can have substantial costs. With the exception of perhaps Zimbabwe, however, most of these countries have relatively small enterprises which do not invest much in training and firm-specific human capital, and operate with relatively simple production technologies (see Tyler Biggs, Manju Shah Shah is a Persian term for a monarch (ruler) that has been adopted in many other languages. This term is a Post Islamic Revolution term for monarchs in Iran which is replaced by valie faghih or Supreme Leader. and Pradeep Srivastava , Technological Capabilities and Learning in African Enterprises, Washington, DC: World Bank, for more details on this subject). Therefore, one would expect a loss to the firm from attrition of experienced workers, but the magnitude of the loss might not be as great as in more advanced countries, where firms devote more resources to building firm-specific capital.
AIDS-Related Worker Attrition and Firm Performance
The characteristics of workers departing firms raises the possibility that observed attrition due to sickness and death affects firm performance, but the ultimate confirmation of such an impact rests on the estimated magnitude of the relationship between attrition and actual enterprise performance indicators. We begin by examining the correlation between overall attrition rates and firm performance. To investigate this relationship, we divide the sample of all firms into two groups, those with overall attrition rates above the median value Noun 1. median value - the value below which 50% of the cases fall
statistics - a branch of applied mathematics concerned with the collection and interpretation of quantitative data and the use of probability theory to estimate population for the full sample and those with attrition rates less than the median. A statistical test is then performed to examine whether there is a significant difference in the mean value of several broad firm performance indicators between the two groups. Table 5 presents the results of this mean-equivalence test. The performance measures used are the firm's profit to sales ratio, average labor productivity and average capital productivity.
As Table 5 indicates, mean value added Value Added
The enhancement a company gives its product or service before offering the product to customers.
This can either increase the products price or value. per unit of capital and labor are both higher for firms with lower rates of attrition, the mean value of the profit to sales ratio shows no significant difference between the two groups. These results suggest that higher rates of labor force attrition have a negative impact (although not much on the current "bottom line") on firm performance; consequently, higher rates of attrition from health-related factors like the AIDS epidemic may adversely affect firms over time.
Taking the analysis a step farther, we use the same mean-equivalence test to investigate mean differences in firm performance, grouping firms based upon whether they have positive or zero attrition due to sickness and death. This grouping is more relevant for testing the current impact of AIDS on firms. Table 6 shows the results of these statistical tests. One can find no discernible dis·cern·i·ble
Perceptible, as by the faculty of vision or the intellect. See Synonyms at perceptible.
dis·cerni·bly adv. difference in mean firm performance between the two groups, except for the capital productivity measure. These results are intuitively appealing, because firm performance is much more likely to be a function (if at all) of the overall attrition rate, than of the smaller subset A group of commands or functions that do not include all the capabilities of the original specification. Software or hardware components designed for the subset will also work with the original. due to sickness and death. However, as the AIDS epidemic progresses, if not controlled, one would expect to see much greater attrition due to sickness and death with probable adverse consequences for firm performance.
Not finding a significant statistical relationship between worker attrition due to sickness and death and firm performance using the mean equivalence test could simply be the result of the crudeness of the test itself. To address this possibility and to control for several factors that might be influencing the empirical relationship In science, an empirical relationship is one based solely on observation rather than theory. An empirical relationship requires only confirmatory data irrespective of theoretical basis. , we utilize multivariable regressions with firm value added as the dependent variable. More specifically, we consider the neoclassical production function in terms of value added per employee as a function of the capital-labor ratio. This basic specification is augmented by the rate of overall work force attrition, and then by attrition due to sickness and death, to analyze their separate impacts on firm performance.
Table 7 presents regressions on value-added per worker with three different specifications for the rate of labor force attrition. The estimation estimation
In mathematics, use of a function or formula to derive a solution or make a prediction. Unlike approximation, it has precise connotations. In statistics, for example, it connotes the careful selection and testing of a function called an estimator. method is ordinary least squares, corrected for heteroscadasticity, and using sector dummies, SEC31-33, and country dummies, GHANAD, ZENYAD, ZIMBWD AND TANZAD, to control for sector-specific and country-specific effects.
In the first specification, the overall rate of labor force attrition, ATTRATE, is treated as an exogenous variable Exogenous variable
A variable whose value is determined outside the model in which it is used. Related: Endogenous variable . The coefficient coefficient /co·ef·fi·cient/ (ko?ah-fish´int)
1. an expression of the change or effect produced by variation in certain factors, or of the ratio between two different quantities.
2. of attrition in this specification, as the estimates in Table 7 indicate, has the wrong sign and is insignificant. In Specification II, the production function is augmented by the overall attrition rate, but in the form of a dummy variable This article is not about "dummy variables" as that term is usually understood in mathematics. See free variables and bound variables.
In regression analysis, a dummy variable , ATTRATD, with the value of one for firms with higher than median attrition and zero for firms with lower than median attrition. In this case, the coefficient of labor force attrition has the correct sign, but is still insignificant. Finally, Specification III considers the possibility that costs of attrition to firms are non-linear, ATTRSQ. That is, at low rates of attrition, the costs are insignificant, but costs increase at higher rates of attrition. Again the results indicate the absence of any significant impact of worker attrition on firm's value-added. The coefficient of attrition indicates a negative effect on value added, at a rate that decreases (in absolute terms (Alg.) such as are known, or which do not contain the unknown quantity.
See also: Absolute ) at higher rates of attrition, but the coefficient is not significant. As might be expected, when the production function is augmented by attrition rates due to sickness and death, rather than the overall attrition rate (results not presented here), similar insignificant results are obtained.
Accounting for Firms' Coping with Attrition
One important reason why we may not find a significant empirical relationship between worker attrition and firm performance is that firms take attrition into account when making decisions. For example, if the costs of attrition are not uniform across firms, firms who suffer higher costs of attrition may take this into account and offer higher wages to induce more workers to stay. Another possibility might be that attrition is completely endogenous endogenous /en·dog·e·nous/ (en-doj´e-nus) produced within or caused by factors within the organism.
1. Originating or produced within an organism, tissue, or cell. . That is, firms' decisions about production and investment in the current period take worker attrition fully into account. To control for these various types of "coping" behavior, we test two additional models of the relationship between worker attrition and value added: an efficiency-wage model and a two-stage least squares regression regression, in psychology: see defense mechanism.
In statistics, a process for determining a line or curve that best represents the general trend of a data set. to control for endogeneity.
Taking the efficiency-wage model first, we estimate the parameters of an equation which assumes that firms suffering higher costs from any level of worker attrition pay wage premiums to induce lower attrition rates. Since the costs to firms of attrition are not observable directly, one can view the wage premium offered by firms as a proxy for the higher costs. To capture this effect, our dummy Sham; make-believe; pretended; imitation. Person who serves in place of another, or who serves until the proper person is named or available to take his place (e.g., dummy corporate directors; dummy owners of real estate). variables grouping firms by high versus low attrition can be interacted with mean wages paid by the firm, ATTRWGE. This is equivalent to specifying the wage premium as equaling the gap between the firm's mean wage and the minimum wage in the economy. Table 8 presents the results of regressions both for overall attrition, ATTRATD, and for attrition due to sickness and death, ATTSICD.
The results indicate that, when we account for differences across firms in the costs of attrition and wage premia, one finds overall attrition and attrition due to sickness and death both appear to adversely affect firm performance (although the coefficient for ATTSICD is not significant). Moreover, this impact on performance increases as the costs of attrition rise: firms that pay higher wages are more likely to be adversely affected by higher attrition than others.
What happens when we assume that labor force attrition is completely endogenous to the firm's current decisions? Particularly for overall work force attrition, it seems reasonable to expect that entrepreneurs have some notion about what average labor attrition might be in the current period and, as a consequence, make operating decisions taking this estimated rate into account. The attrition due to sickness and death component of overall attrition is probably more difficult for entrepreneurs to estimate. However, discussions with entrepreneurs during the course of our interviews indicated that, in the severely HIV-infected countries, like Zimbabwe and Zambia, firm managers openly discuss observed problems with AIDS in the work force. Managers generally can tell you the percentage of the work force that appears to show signs of HIV illness. Also, because most firms have funeral expenses as part of their normal labor contracts, managers can tell you how much these costs are rising on average. In many cases, entrepreneurs stated that they are planning for increased attrition due to AIDS, but felt that, in the current economic situation of severe unemployment, replacement of unskilled workers will not be difficult. Hence, there are reasons to believe that attrition due to AIDS may also be endogenous to current firm decisions.
To control for the possibility of endogeneity, we use a two-stage least squares procedure to examine the relationship between attrition and value added. In the first stage, the endogenous variable Endogenous variable
A value determined within the context of a model. Related: Exogenous variable. , overall attrition, is regressed on a set of exogenous Exogenous
Describes facts outside the control of the firm. Converse of endogenous. , predetermined variables Predetermined variables are variables that were determined prior to the current period. In econometric models this implies that the current period error term is uncorrelated with current and lagged values of the predetermined variable but may be correlated with future values. chosen from the previous year of survey interviews. The fitted value of this variable is then used as an exogenous regressor in the second stage to examine its effect on value added. In this second stage, we also control for the possibility that costs of attrition are not uniform across firms and resulting wage premia.
Table 9 presents the first and second stage results of the regressions. In stage 1, current overall attrition, ATTRATE, is regressed on ATTRAT92, the previous year's attrition rate; PROFSAL92, the previous year's profit/sales ratio, and firm size in the previous year, measured as the log of the number of workers, LOGWK92. Results indicate that overall attrition is positively correlated cor·re·late
v. cor·re·lat·ed, cor·re·lat·ing, cor·re·lates
1. To put or bring into causal, complementary, parallel, or reciprocal relation.
2. with prior year attrition and negatively correlated with firm size, as we noted earlier in this paper. The fitted value of ATTRATE obtained from this first stage is then used in the second stage to examine the impact of attrition on firm performance. The second stage results, controlling for endogeneity, indicate different results than those obtained by ordinary least squares. Overall attrition now has a significant negative impact on value added. The interaction term, ATTWGE, is also significant, suggesting that firms are also taking account of the fact that costs of attrition are not uniform across firms and trying to offset these costs by paying wage premia.
Thus, it would appear that African firms do take worker attrition into account in current operational decisions and that attrition does influence firm performance. The effect of attrition due to sickness and death, however, is not so clear. The results of a similar regression controlling for endogeneity of attrition due to sickness and death are shown in Table 10. The attrition variable, ATTSICR, has the correct sign but is not significant. Hence, we are left with these conclusions. The evidence points to a significant influence of worker attrition on firm performance. The AIDS epidemic, however, because it has had, to date, only a minor effect on total attrition, is yet to have a significant impact on firms.
The effect of the AIDS epidemic on firms depends upon the strength of two factors. First, work force attrition resulting from illness or death due to HIV infection must be a large proportion of the total worker attrition in firms. Second, the higher rates of worker attrition must adversely affect firms' costs and performance.
Using primary survey data from five sub-Saharan African countries, covering 992 firms with total employment of 115,136 workers, this study finds both effects to be minor. Although the observed attrition rates due to death or sickness of workers conform to Verb 1. conform to - satisfy a condition or restriction; "Does this paper meet the requirements for the degree?"
coordinate - be co-ordinated; "These activities coordinate well" patterns of HIV seroprevalence across the countries, the quantitative impact of AIDS-related attrition seems quite small. Even under the extreme assumption that all attrition due to sickness or death observed in the sample is AIDS related, the attrition rates are lower than would be suggested by HIV seroprevalence data. In addition, attrition due to illness or death constitutes a relatively small proportion of total work force attrition in these countries, equaling about 3% in Ghana and Tanzania and 12-13% in the other countries.
The data also show that the costs to firms of replacing departed workers increase in proportion to the skill levels but are, on the whole, not very high. Replacement for all nonprofessional non·pro·fes·sion·al
One who is not a professional.
nonpro·fes workers are found in less than three weeks on average while it takes slightly less than six months to replace professional vacancies. However, observed rates of worker attrition do appear to influence firm performance. When we control for firms taking worker attrition into account in their operational decisions, attrition significantly affects performance. AIDS-related attrition, however, does not.
One is left, then, with the overall conclusion that AIDS has not yet had a significantly negative effect on African firms. But, left unabated, and given current trends in the HIV seroprevalence data for many countries, this finding may not hold true for long.
* We thank Mead Over and the participants at the "AIDS and Development: The Role of Government" workshop in Brussels in May 1996 for helpful comments on this paper.
(1) Rates of attrition due to sickness or death are not available for 1993.
(2) See Table 6-1, p. 224 of National Research Council's 1996 publication. Preventing and Mitigating AIDS in Sub- Saharan Africa: Research and Data Priorities for the Social and Behavioral Sciences behavioral sciences,
n.pl those sciences devoted to the study of human and animal behavior. , Washington, DC: National Academy Press for a more complete explanation of the assumptions behind the calculation of these estimates.
(3) The attrition rates due to sickness and death found in our sample are very close to those found in detailed case studies of firms in high seroprevalence countries. For example, case studies in Zambia--carried out by Smith and Whiteside of Cambridge University Cambridge University, at Cambridge, England, one of the oldest English-language universities in the world. Originating in the early 12th cent. (legend places its origin even earlier than that of Oxford Univ. , England, and University of Natal The University of Natal was a university in Natal, and later KwaZulu-Natal in South Africa. It was founded in 1910 as the Natal University College in Pietermaritzburg, and expanded to include a campus in Durban in 1931. , South Africa--found that attrition due to sickness and death averaged about two to three percent of the work force in the 1993-95 period. The AIDS-related component of this attrition was found to be about 70 percent of the cases. Moreover, they found in each of the three large enterprises studied that AIDS-related attrition was on the rise and productivity effects (losses in work hours) and medical costs in these firms were also increasing. Their overall conclusion is that, "Historical records of mortality and morbidity morbidity /mor·bid·i·ty/ (mor-bid´it-e)
1. a diseased condition or state.
2. the incidence or prevalence of a disease or of all diseases in a population.
n. [in the firms studied] suggest that the HIV/AIDS epidemic is yet to have a major impact [on firms] although the upward trend in indicators is very distinct." (James Smith James Smith is the name of: People named James Smith
CDC - Control Data Corporation ." London: Commonwealth Development Corporation).
Table 1 Worker Attrition in Ghana, Kenya, Zimbabwe, Zambia and Tanzania: Total and By Sickness or Death Rates of attrition due to sickness or death are not available for 1993. Total in Sample Percentage HIV+ Country Low Risk Firms Workers Zambia 24.7 194 14,582 Zimbabwe 20.5 199 59,210 Kenya 17.1 214 17,126 Tanzania 16.1 197 14,611 Ghana 2.2 188 9,607 Total 992 115,136 Atrrition due to Attrition due to All Causes Sickness or Death Percentage Percentage of all of all Country Number workers Total workers Zambia 3,035 20.8 (25) 360 2.5 Zimbabwe 5,366 9.1 (14) 695 1.2 Kenya 1,325 7.7 (6) 151 0.9 Tanzania 2826 19.3 (8) 89 0.6 Ghana 1,110 11.6 (9) 30 0.3 Total 13,662 11.9 1,325 1.15 Note: All figures are for 1994, except for those in parentheses, which are from 1993. Source: Seroprevalence data from: U.S. Bureau of Census (1994), Research Paper #17, Washington, DC. Other data from RPED Panel Survey. Table 2 Rate Of Attrition By Firm Size Class: Mean Attrition Rates Ghana Kenya Zimbabwe Attr. Attr. Attr. Rate Rate Rate Due Due Due Avg. To Avg. To Avg. To Firm Attr. Sick/ Attr. Sick/ Attr. Sick/ Size Rate Death Rate Death Rate Death Micro 22.76 .34 26.3 4.7 39.2 .33 Small 15.7 .61 10.0 1.3 18.6 1.2 Med. 6.3 .49 5.2 .89 7.3 1.5 Large 10.7 .17 7.3 .71 9.0 1.2 Tanzania Zambia Attr. Attr. Rate Rate Due Due Avg. To Avg. To Firm Attr. Sick/ Attr. Sick/ Size Rate Death Rate Death Micro 8.2 .29 28.5 2.6 Small 12.1 .77 34.4 1.9 Med. 11.8 .65 21.3 1.6 Large 21.5 .59 18.5 2.7 Note: Rate calculated as number of workers leaving divided by total number of workers in that size class. Source: RPED Panel Survey, 1994 Table 3 Relationship of Skill Level to Ease of Replacement of a Deceased Worker Skill Number of Found a Still Category Observations Replacement Looking Professional 16 4 6 (25%) (37.5%) Skilled 89 47 12 (53%) (13%) Operator 124 61 8 (49%) (6.5%) Unskilled 210 94 8 (45%) (4%) Total 439 206 34 Average Weeks 3 Decided to Find a Not to Replacement Skill Replace for Those Category Employee Found Professional 6 23.8 (37.5%) Skilled 30 3.0 (34%) Operator 55 3.1 (44.5%) Unskilled 108 1.9 (51%) Total 199 Source: RPED Panel Survey, 1994 Table 4 Average Age at Death, Average Years of Experience of Deceased Workers, by Skill Category Mean Age in Skill Number of Years at Median age Category Observations Death (std) at death Professional 10 47.5 45 (10.0) Skilled 87 35.7 35 (12.5) Operator 113 40.1 39 (11.4) Unskilled 182 39.7 40 (11.3) Median Mean Years Years Skill Worked at Worked at Category Firm (std) Firm Professional 12.9 11 (8.9) Skilled 9.7 10 (6.9) Operator 10.9 10 (7.9) Unskilled 9.1 7 (7.1) Source: RPED Panel Survey, 1994 Table 5 Performance of Firms with High versus Low Attrition Rates Firms are divided into two groups: ATTRATD=1 for firm's with attrition rates lying above the median; ATTRATD=0 otherwise. ATTRATD=1 ATTRATD=0 T-statistic Profit to Total .13 .14 0.53 Sales Ratio (.16) (.16) Value Added per 4007.7 4967.8 1.65 ** Unit of Labor (4997) (8092) Value Added per 1.86 3.23 2.17 * Unit of Capital (4.35) (8.64) Average labor 1129 1110 -.17 costs (1267) (1243) Figures in parentheses are standard errors. * Significant at 5% level ** Significant at 10% level Table 6 Mean Difference Test: Performance of Firms with High versus Low Attrition Rates Due to Sickness or Death Firms are divided into two groups: ATTSICD=1 for firms with attrition rates from sickness and death above the median value, ATTSICD=2 ATTSICD=1 ATTSICD=0 T STAT Profit to total .14 .14 -.13 sales ratio (.15) (.17) value added per 5191.3 4251.3 -1.59 unit of labor (5806.3) (6813.1) value added per 1.52 2.82 1.98 * unit of capital (4.62) (7.22) Figures in parentheses are standard errors. * Significant at 5% level ** Significant at 10% level Table 7 Regressions: Augmented Production Function with Attrition Rate Spec I Spec II Spec III INTERCEP 3.57 * 3.66 * 3.61 * (0.4895) (0.487) (0.490) LCAPLAB 0.287 * 0.289 * 0.287 * (0.0321) (0.032) (0.032) ATTRATE 0.0885 -0.139 -0.114 ATTRATD (0.124) (0.107) (0.229) ATTRSQ 0.067 (0.064) SEC31 0.305 * 0.280 * 0.306 * (0.149) (0.1489) (0.1490) SEC32 -0.1322 -0.133 -0.121 (0.143) (0.143) (0.1435) SEC33 -0.2072 -0.21 -0.200 (0.1544) (0.15) (0.1545) GHANAD -0.715 * -0.74 * -0.723 * (0.16) (0.162) (0.1623) KENYAD 0.395 * 0.35 * 0.380 * (0.17) (0.172) (0.1738) ZIMBWD 1.78 * 1.76 * 1.759 * (0.19) (0.193) (0.1952) TANZAD -0.354 * -0.41 * -0.371 ** (0.197) (0.19) (0.1982) ADJ. RSQ .26 .26 .26 Figures in parentheses are standard errors. * Significant at 5% level ** Significant at 10% level Table 8 Regressions: Augmented Production Function with Attrition Rate Interacted with Mean Firm Wage Spec I: Overall Spec II: Attrition ATTSICR INTERCEPT 4.01 * 3.80 * (0.48) (.49) LCAPLAB 0.27 * .27 * (0.03) (.03) ATTRATD -.49 * (.12) ATTRWGE .0003 * (.12) ATTSICD -.25 (.18) ATTSWGE .0003 * (.0001) SEC31 .27 * .29 * (.14) (.15) SEC32 -0.08 -.08 (0.59) (.14) SEC33 -.17 -.17 (.15) (.15) GHANAD -.70 * -.68 * (.16) (.16) KENYAD .35 * .38 * (.17) (.17) ZIMBWD 1.46 * 1.62 * (.20) (.20) TANZAD -.42 * -.36 ** (.19) (.19) RSQ .30 .27 Figures in parentheses are standard errors. * Significant at 5% level ** Significant at 10% level Table 9 2SLS results : Impact of Attrition on Value Added First Second Stage Stage INTERCEPT -0.11 4.90 * (.22) (.86) ATTRAT92 .29 ** (.17) PROFSAL92 .23 (.19) LOGWK92 -.01 (.03) LCAPLAB .26 * (.04) ATTRATE -2.15 * (1.03) ATTWAGE .0006 * (.0001) SEC31 .29 ** (.15) SEC32 -.04 (.14) SEC33 -.14 (.16) GHANAD -.77 * (.18) KENYAD .36 ** (.19) ZIMBWD 1.26 * (.22) TANZAD -.28 (.21) RSQ .39 Figures in parentheses are standard errors. * Significant at 5% level ** Significant at 10% level Table 10 2SLS RESULTS: Impact of Attrition Due to Sickness and Death on Value Added First Stage Second Stage INTERCEPT -2.24 4.28 * (.28) (.52) P92 .044 * (.012) P92m .85 * (.27) ATTSIR92 .011 (.29) LOGWK92 -.21 * (.04) LCAPLAB .22 * (.04) ATTSICR -.05 (.49) ATTWAGE .0009 * (.0002) SEC31 .19 (.15) SEC32 -.10 (.15) SEC33 -.14 (.16) GHANAD -.65 * (.18) KENYAD .41 * (.18) ZIMBWD 1.11 * (.23) TANZAD -.28 (.20) RSQ 0.35 Figures in parentheses are standard errors. * Significant at 5% level ** Significant at 10% level