2.1 Measuring Product Quality

The earliest attempts to measure quality wereby Reference FeenstraFeenstra (1994) and, subsequently, Reference Broda and WeinsteinBroda and Weinstein (2006). Using constant elasticity of substitution (CES)–type utility functions, Reference FeenstraFeenstra (1994) constructed a price index allowing for different sets of product varieties and quality variations in them over time. However, in a demand and supply equilibrium, if the new varieties are not taken into account it results in a bias; the extent of the bias depends on the elasticity of substitution between all the varieties. Reference FeenstraFeenstra’s (1994) solution was to estimate the elasticity of substitution between varieties from each country using the generalised method of moments.Footnote ² In contrast, Reference SchottSchott (2004), Reference Hummels and KlenowHummels and Klenow (2005), Reference HallakHallak (2006) and, more recently, Reference Fan, Li and YeapleFan, Li and Yeaple (2018) use as a proxy for product quality the unit value, i.e. the observable average trade price for each product category. The idea here is that higher-quality goods sell at higher prices and higher price signals higher quality. However, unit values are noisy proxies as they are driven by a series of other factors, including production cost differences such as wage differentials. Moreover, changes in unit values over time may reflect changes in quality-adjusted prices (owing to supply or demand shocks), rather than changes in quality itself. So unit-value-equivalent quality estimates fail to differentiate across the vertical (e.g., comfort) and horizontal (e.g., style) attributes that products possess. This is what Reference Amiti and KhandelwalAmiti and Khandelwal (2009) and Reference KhandelwalKhandelwal (2010) built upon. In their measure of quality, when two products have identical unit values, the product with the higher market share is assigned higher quality, the amount depending on the slope of the demand schedule. To estimate quality, they use a nested logit demand framework based on Reference BerryBerry (1994), where they define quality as the vertical component of the model and assign astructural definition to it as the mean valuation that US consumers attach to an imported product. In a discrete-choice framework, higher quality identified with higher market share will act as a parallel demand shifter. However, what they missed out is that other factors such as changes in tastes will also affect export market shares. That is, there may be shifts in the demand curve not induced by a change in export quality. As pointed out by Reference VandenbusscheVandenbussche (2014), a product of a certain quality exported by a country could have different market shares in two destination countries due to the differences in preferences among the consumers in the two markets. So, taking into account additional demand-shifting parameters gives a more unbiased measure of quality. When there is a change in the market share of two different varieties of a product in the destination country, one can thus differentiate whether it is due to a change in the quality (if there is a vertical demand shift) or to a change in tastes (if there is a slope shift).Footnote ³

Reference VandenbusscheVandenbussche (2014) also criticised the earlier CES approach of Reference FeenstraFeenstra (1994) by arguing that CES across varieties fails to distinguish between vertical and horizontal differentiation. Despite introducing a firm-product-specific demand shock that accounts for sales variation of the same firm-product across countries without affecting prices (Reference Bernard, Redding and SchottBernard, Redding and Schott, 2011), this issue remains unresolved. So a well-defined set of consumer preferences is necessary to clearly differentiate quality differences from taste differences across product varieties. Di Comite, Thisse and Reference VandenbusscheVandenbussche (2014) came up with a clearer approach in which they tried to disentangle the interplay of horizontal and vertical differentiation to infer which shifts in demand are actually attributable to changes in quality. They argued that changes in taste will induce variations in the quantities of the variety demanded but not affect the willingness to pay. Using an extended Reference Melitz and OttavianoMelitz–Ottaviano (2008) model and Belgian firm-product data, they considered quality differences between firms as firm-specific parallel demand shocks, in addition to productivity differences, that determine firms’ export market performance. They generated an indicator for unobservable quality using export prices (unit values) and product-level mark-ups created from firm-level data on variable input costs and sales. They also captured competition effects in the destination market as the consumption of all substitute products available to consumers there. However, as they could not meet the data requirements to assign a quality level to each product, the quality measure obtained by them is only a relative quality ranking of each product as compared to other competitors of the same product in Europe.

Other strategies for quality estimation looked at addressing country-specific issues for short periods of time and few product varieties. For example, Reference KhandelwalKhandelwal (2010) analysed the effect of import competition on quality upgrading in the United States, mainly aiming to establish that low wage competition in the United States causes bigger losses in employment and output levels for US sectors with short quality ladders. Following Reference KhandelwalKhandelwal (2010) and Reference Hallak and SivadasanHallak and Sivadasan (2013), Reference Hu, Parsely and TanHu, Parsely and Tan (2017) defined quality as any attribute other than price that raises consumers’ demand, to examine the effect of appreciation of domestic currency relative to that of the source country on exported product quality using Chinese Customs data between 2000 and 2006. Similarly, while examining the effects of changes in real exchange rates using Argentinean firm-level wine export values and volumes between 2002 and 2009, Reference Chen and JuvenalChen and Juvenal (2014) took wine ratings by two global rating systems as a measure of quality. To address the potential endogeneity of quality in explaining unit values and export volumes, they further used appropriate instruments for quality based on geography and weather-related factors.

Another novel approach was taken up more recently by Reference Piveteau and SmagghuePiveteau and Smagghue (2019), who attempted to identify quality from the demand side at the firm-product-destination-year level. Similar to Reference KhandelwalKhandelwal (2010), they used a CES demand system but identified the quality of a product as a utility shifter, which implies that it is variations in sales and not price movements that explain export quality variations over time and across firms. They presented a new instrument for obtaining the price of firms’ exports; it works by interacting firm-specific importing shares with real exchange rates and then identifying firm-level quality from residual export variations, after controlling for prices. They argue that this instrument is exogenous to both the quality choices that firms make and measurement errors on prices, which constitutes an improvement relative to the existing instruments in the literature, thereby providing consistent estimates of the demand functions using trade data. However, the methodology was applied only to French firms for a short period, from 1997 to 2010; such country-specific estimation is not suitable for making cross-country comparisons, especially those incorporating developing countries.

Reference Henn, Papageorgiou and SpataforaHenn, Papageorgiou and Spatafora (2013) fill this gap with a parallel endeavour to develop new estimates of quality taken up under the World Trade Organization (WTO) Economic Research and Statistics Division, as an International Monetary Fund–Department for Internatinal Development (IMF–DFID) research collaboration. Extending the UN–National Bureau of Economic Research (NBER) dataset, they provide us with the most extensive quality indices for 200 countries for the period 1962–2014, and covering 851 products at the United Nations (UN) Standard International Trade Classificaiton (SITC) four-digit level. The trade dataset is constructed by supplementing importer-reported data with exporter-reported data where the former does not exist. The dataset contains 45.3 million observations on bilateral trade values and quantities at the SITC four-digit level. The estimation methodology derives quality from unit values, but with two important adjustments. The methodology is a modified version of Reference HallakHallak (2006). As a first step, it determines the trade price (equivalently, the unit value) for any given product. Prices reflect three factors: unobservable quality, per capita exporter income and distance between exporter and importer. This accounts for selection bias. Typically, the composition of exports to more distant destinations is tilted towards higher-priced goods because of higher shipping costs. Next, a quality-augmented gravity equation is specified, separately for each product, because preferences for quality and trade costs may vary across products. The estimation equation is then obtained by substituting observables for the unobservable quality parameter in the gravity equation. It is estimated separately for each of the 851 product categories at the SITC four-digit level. Regression coefficients are used to calculate a comprehensive set of quality estimates which are then aggregated across all importers, using current trade values as weights, and then to SITC three-digit, two-digit, one-digit and country-level aggregates. Such an extensive and detailed methodology significantly contributes to bridging the large gap that the existing methodologies have left behind. It provides a set of quality estimates with country, product and time coverage that is as wide as possible, which not only makes country-specific empirical studies comparable in terms of unified quality estimates but also creates scope for examining the underlying causal factors in a cross-country study.

2.2 Cross-Country Quality Dispersions

Reference Henn, Papageorgiou and SpataforaHenn et al. (2013) identify significant cross-country heterogeneity in the growth rate of quality. Stylised facts put forth suggest that reducing barriers to entry into new sectors can allow economies to benefit from rapid quality convergence over time. Their findings that quality upgrading is particularly rapid during the early stages of development, with the process largely completed as a country reaches upper-middle-income status, is a learning lesson for developing countries. While low-income countries suffer from the poor quality of their export products, a similar low-export-quality phenomenon can be observed for many lower-middle- and upper-middle-income countries as well.

Figure 1 reflects upon this by comparing the export quality indices of manufactured goods of some sample countries from different income subgroups with those of the United States for the period 1963–2010.Footnote ⁴ Note that the quality levels of primary goods such as mining extractions, tea, coffee, rice, etc., are often reflections of climatic and other natural causes, which may not be changed through subsequent processing.Footnote ⁵ Thus, comparisons of quality across developed and developing countries are relevant for manufactured goods for which there is scope for substantial transformations of the attributes of the basic inputs. If we look at the range of quality indices in the four panels in Figure 1, it clearly brings out the stark gap in the quality of manufactured goods between the rich and developed North and the less-developed South. Panel (a) reveals that the quality indices for Australia, Japan and the UK were at par with those of the United States in the 1960s and early 1970s, but thereafter, while Japan produced marginally better-quality manufactured goods than the United States, Australia and the UK fell behind. On the other hand, for low-income countries such as the four African countries reported in panel (d), as well as for lower-middle-income countries such as India, Indonesia, Morocco and Sri Lanka, as reported in panel (c), lower average quality of manufactured exports is evident. Of course, low-income countries lag behind to a larger extent than lower-middle-income countries. However, the striking similarity among these four low-income countries is that the quality levels of their manufactured goods have worsened steadily since the late 1970s. In contrast, India, Indonesia and Sri Lanka have improved the quality of their manufactured goods since the late 1970s, with quality upgrading being most spectacular in Indonesia, although the average quality of the manufactured exports of these countries remains well below that of the United States and other rich countries. An interesting observation that we can make here is that Uganda produced better-quality manufactured goods, on average, than all these four lower-middle-income countries during the 1960s and 1970s. Lower average quality for the four upper-middle-income countries – Brazil, China, Mexico and South Africa – is reported in panel (b). The quality of Brazil’s manufactured goods deteriorated steadily after the 1960s and by 2008 had fallen to the level of quality produced by India, Indonesia and Sri Lanka. China has achieved remarkable improvement since the early 1990s, catching up with South Africa by the late 1990s and with Mexico by 2005.

Source: Authors’ calculation based on IMF data, 2014 (www.imf.org/external/np/res/dfidimf/diversification.htm)

Figure 1 Quality indices of manufactured goods relative to those of the USA (1963–2010).

2.3 Asymmetric Quality Variations

In addition to the non-monotonic movements in average quality among the developing countries presented in Figure 1, a deeper look into the data also reveals asymmetric variations in quality across some export product groups for some of the developing countries. A first-hand and somewhat crude indicator of asymmetric quality variations is the pair-wise correlation value. A positive correlation value for a pair of products during a particular time interval suggests that their quality levels have moved mostly in tandem (or in the same direction). A negative correlation value, on the other hand, indicates that quality variations have been asymmetric for the concerned pair of product groups.

The correlation tables below exemplify the quality variations across five product groups for several countries from different income groups. In line with the technological explanation of Reference Ganguly and AcharyyaGanguly and Acharyya (2021), the five product groups (at the SITC two-digit level) considered for each country are a sample of product groups with different factor intensity requirements for their respective quality upgrading. For example, while higher-quality medicinal and pharmaceutical products (MPP) and chemical elements and compounds (CC) are relatively more skill intensive than they are capital intensive, higher-quality transport equipment (TE) and electrical machinery, appliances and apparatus (EM) require more capital than skilled labour. On the other hand, petroleum and petroleum products (PPP) are largely imported input intensive in most countries.

Tables 1 and 2 report correlations in quality across these five product groups for Japan and the United States, respectively, calculated for the period 1980–2000. While significant negative correlation values can be observed only between product groups PPP and MPP, and between TE and MPP for the United States, such asymmetric variations are completely absent in the case of Japan. Similar absence (or very few cases) of asymmetric variations has been recorded for other high-income countries such as the UK and Australia. This is indicative of the link between the level of development of a country and its gradual attainment of quality convergence, which was also pointed out by Reference Henn, Papageorgiou and SpataforaHenn et al. (2013).

Table 3 shows the case of a high-middle-income country, Brazil. In the period for which the correlation values are calculated, asymmetric quality variations across product groups are found to be more pronounced and significant. Further, TE has moved in the opposite direction with respect to all four of the other categories. It is interesting to note that this period of study also coincides with the commencement of the 1980 Debt Crisis in Latin America. The lost decade of development for the debt-ridden countries therein is reflected in highly diverging and irregular quality movements across product groups for Mexico and Argentina as well.

However, China’s situation was quite different from that of the other high-middle-income countries, as mentioned earlier. Positive and significant correlation values can be observed for all pairs of product groups in Table 4 from the mid-1980s till 2010, indicating no dissimilar cross-product quality variations whatsoever. This is also the same timeline during which China started pushing up its aggregate quality levels to catch up with high-quality-producing countries like the United States (see Figure 1, panel (b)).

As one goes down the ladder of development, coming to the lower-middle-income countries, evidence of asymmetric quality variations varies across countries. Line graphs for Indonesia (Figure 2) and the Philippines (Figure 3) show that while both were similarly and adversely hit by the financial contagion in Asia in the late 1990s, significant asymmetric quality variations are present in the former country but almost absent from the latter.

Figure 2 Quality variations: Indonesia.

Source: Authors’ calculation based on IMF data, 2013

Figure 3 Quality variations: Philippines.

Source: Authors’ calculation based on IMF data, 2013

The trends mapped here show that quality levels of CC, TE and MPP move opposite to each other during almost the entire period of study for Indonesia with statistically significant negative correlation values: -0.5557 between CC and MPP, and -0.3939 between CC and TE. On the other hand, significant positive correlation between these very product groups – 0.4739 between CC and MPP, 0.4424 between TE and CC – is observed for the Philippines, also captured through quality lines that move in tandem with each other. All of this suggests that the trade, globalisation and development policies adopted by these countries during 1985–2010 had asymmetric effects on the quality of export goods across different product categories.

In the case of India, another developing low-middle-income country, trends in asymmetric quality variations have been observed across a different set of product groups. Such asymmetric variations are more pronounced after the mid-1980s, which marks the beginning of the liberalisation of import of capital and intermediate goods by India. As documented in Reference Ganguly, Acharyya, Bagli, Chakrabarti and GuhaGanguly and Acharyya (2022a), significant negative correlation between EM and non-ferrous metals (-0.5944), between non-ferrous metals and MPP (-0.7302), and between leather manufactures and plastic materials (-0.7447) could be observed. Similar trends have been observed with leather manufactures, PPP, MPP, CC, and EM during that period.

Finally, coming to the low-income countries, Uganda, being one of the largest merchandise exporters to high-income countries (per the World Development Indicators 2013 Bank database), exhibits even more profound asymmetric quality variation among these product groups (see Table 5). Extending the period of analysis by a further ten years retains this dimension of asymmetric quality variations.

These observations suggest that trade liberalisation policies adopted by countries at different points during the wave of globalisation may have affected the quality of their export goods asymmetrically. Export-promotion strategies such as tariff reductions raise the prices of some domestic inputs while lowing the prices of a few others. This may have differential effects on the choice of quality of export products, which varies from one product to another with respect to the skill or capital intensities required for higher-quality varieties.

To conclude, quality estimates developed by Reference Henn, Papageorgiou and SpataforaHenn et al. (2013) fill in the gap in the quality estimation literature where cross-country comparisons, especially those incorporating developing countries, could not be made using country-specific estimation procedures in different studies. Using such estimates, we document the stark gap in the quality of manufactured goods between the rich and developed North and the less-developed South, as well as asymmetric quality variations across different product categories.

3.1 Demand-Side Explanations

Demand-side explanations for low product quality date back to the three pioneering analyses of quality choices by firms in a partial equilibrium framework by Reference Mussa and RosenMussa and Rosen (1978), Reference Gabszewicz and ThisseGabszewicz and Thisse (1979) and Reference Shaked and SuttonShaked and Sutton (1982). A substantial literature followed these groundbreaking works that demonstrated quality discrimination and under-provision of quality by a monopolist as well as quality differentiation by oligopolistic firms when consumers are heterogeneous with respect to their tastes or incomes. There is also a small and relatively newer general equilibrium analysis relating income disparities and distribution to quality from the demand side (Reference Fajgelbaum, Grossman and HelpmanFajgelbaum, Grossman and Helpman, 2011). In Sections 3.1.1 and 3.1.2, we briefly discuss these two literatures and their implications for the low-export-quality phenomenon in developing countries.

3.1.1 Income Disparities and Quality Variations: Partial Equilibrium Analysis

The central theme of the partial equilibrium endogenous quality choice models is that when buyers can observe quality a priori, such as in the case of search goods, their marginal willingness to pay (MWP) increases with the level of quality of the good under consideration. This, along with the marginal cost of production increasing with the quality level, determines the profit-maximising quality provided by firms from a set of technologically feasible quality levels. So, for any given cost of quality, if MWP in a market is low, the quality offered will be low as well. Low incomes of consumers in developing countries can also cause their purchasing power (PP) constraint to be binding and thus can cause firms to provide a lower quality than they would otherwise (Reference AcharyyaAcharyya, 2005). Thus, cross-country income differences can explain the low-export-quality phenomenon in developing counties, with or without a (binding) PP constraint. Furthermore, when consumers are heterogeneous, firms discriminate among them by offering different pairs of prices and qualiy levels. This can result in further quality degrading in the developing country.

These implications of cross-country as well as within-country income and demand disparities for the low-export-quality phenomenon of exports by low-income developing countries can be explained as follows. Let $Q\in [0,1]$ denote the index of observable quality. A consumer buys, if at all, only one unit of the good, which is a standard assumption employed in the literature intended to separate out quality and quantity dimensions of demand.Footnote ⁶ This also enables us to focus solely on a consumer’s choice regarding whether to buy a quality-differentiated good, and, if so, which particular quality to buy, from among the different qualities of the good available in the market. Such a decision over the quality dimension of a good – that is, whether to buy and which quality to buy, instead of how much to buy – is driven by the marginal utility (MU) that a consumer derives from a particular level of quality Q net of the unit price that she pays.Footnote ⁷ Higher utility from consuming a better-quality good makes her MWP increase with the quality of the good. The MWP also depends on the type of the consumer. If she is a high type – having either better taste or higher income than another buyer – then she derives larger additional utility from consuming a better quality than other (low-type) buyers and, therefore, is willing to pay more for it. However, to begin with, suppose all buyers in each country are identical in all respects, so that all consumers derive the same MU from a better quality of the good under consideration and thus pay the same additional price for it.

Suppose in each country the good is produced by perfectly competitive firms. Due to higher MWP by a (representative) consumer for a better quality, each firm can get a larger revenue by offering a better quality. But producing a higher quality entails larger costs – both total and marginal costs. Under the simplifying assumption that the marginal cost (MC) of production is invariant with respect to the number of units of the good being produced, although it is larger for a better quality, the profit-maximising quality that each competitive firm will offer will be the one for which the MWP equals the MC.

Now, to focus on how the cross-country disparity in (per capita) income levels explains the low-export-quality phenomenon in a developing (or poor) country, let the MWP depend on the income level, as in Reference Gabszewicz and ThisseGabszewicz and Thisse (1979), for example. Let y denote the income level of a representative consumer in the developing country. Let P be the price of the good in the local market. Then, the net utility V derived by the representative consumer from consuming the good of quality Q at the price P is given by:

$V(y,Q,P)=U(y,Q)-P,U_y>0,U_Q>0,U_{QQ}<0$(1)

where U represents the gross utility. The first partial derivative in Equation (1) indicates that the representative consumer’s utility increases with her income level (or her type) for any given Q; the other two partial derivatives indicate that her utility increases with the quality of the good at a decreasing rate, given her type. Essentially, the MU from quality variation, U_Q, is her MWP for quality. An important property of the net utility function in Equation (1) deserves attention: it is additively separable in quality and price. This makes the upward-sloping iso-net-utility curves in the (Q, P) space, or the indifference curves between price and quality, vertically parallel to each other. This has some far-reaching implications, as we will discuss later. Given that her reservation utility is zero, the consumer buys the good if and only if her net utility $V(\alpha ,Q,P)$ is non-negative.Footnote ⁸ This defines her individually rational (IR) or the market participation (MP) constraint.

Let there be a richer developed country with all consumers being identical there as well. They have the same utility functions and their purchase decisions are also the same as the consumers in the developing country, but they have a higher (per capita) income, y*. By the specification of the gross utility function in Equation (1), therefore, the consumers in the low-income developing country derive lower total as well as marginal utilities than the consumers in the richer country:

$U(y,Q)<U(y^{*},Q),U_Q(y,Q)<U_Q(y^{*},Q)$(2)

Suppose competitive firms in both the countries are identical in terms of technology – including knowledge about the feasible set of qualities – and costs. Let the MC of producing the good, c, be invariant with respect to the output level but increasing in the quality level at an increasing rate:

$c=c(Q),c_Q>0,c_{QQ}>0$(3)

It is then straightforward to check that, under autarchy, quality levels $Q_C$ and $Q_C^{*}$ will be provided by the competitive firms in the low-income country and the rich country, respectively, such that

$U_Q(y,Q_C)=c_Q(Q_C),U_Q(y^{*},Q_C^{*})=c_Q(Q_C^{*})$(4)

By the property of the utility function and that of the cost function, it immediately follows that Q_C < $Q_C^{*}$ . Thus, low (per capita) income explains the low-export-quality phenomenon in the developing country. This is illustrated in Figure 4. The upward-sloping concave curves represent the indifference curves between price and quality for representative consumers in the two countries: The steeper (broken) curve is that for a consumer in the rich country and the flatter (solid) curve is that for a consumer in the low-income developing country. Curves labelled V^* = 0 and V = 0 represent the IR constraints. The upward-sloping convex curve represents the identical MC function specified in Equation (3). The first-order conditions in Equation (4) are satisfied at $E_C$ and $E_C^{*}$ . The net utilities derived by the consumers in the two countries are indicated by the curves labelled $V_C$ and $V_C^{*}$ . These curves being strictly below the IR constraints means that the consumers in each country enjoy strictly positive surpluses at the competitive (autarchic) equilibrium.

Figure 4 Cross-country quality variations

Low income of consumers in developing countries may also imply that their purchasing power (PP) constraint is binding in the sense that some technologically feasible high-quality varieties, if offered in the market, may not be affordable to them because of the high prices of those varieties. If such a binding PP constraint does not allow the consumers to purchase even the competitive quality Q_C, then firms will provide an even lower quality. To illustrate in the simplest possible manner, let a representative consumer in the developing country spend β fraction of her income on the quality-differentiated good and the remainder on the other goods. If β is low, or y is very low, it may be possible that βy < $P_C=c(Q_C)$ , in which case the PP constraint is binding. So this may further lower the quality level offered by firms in the developing country.Footnote ⁹ This is illustrated in Figure 5. The MP constraint is now the kinked curve OMM′. For all $Q<\tilde{Q}$ the IR constraint is binding and thus represents the MP constraint. For all other higher qualities, the PP constraint is binding and thus determines the decision to participate in the market and purchase the good. The competitive firms will now offer the quality ${\overline{Q}}_C$ , which is lower than the Q_C that they would have offered without the binding PP constraint.

Figure 5 Purchasing power constraint

How do the results change when income disparities exist among consumers in each country? By the property of the utility function specified earlier, the MU from quality in each country will be higher for a richer consumer and thus she will be willing to pay a higher price than a poorer consumer in her country for a better quality. This heterogeneity enables firms, competitive or otherwise, to offer different price–quality menus to different consumers if the technologically feasible set of qualities is known to all firms without any additional costs.Footnote ¹⁰ Thus, in each country more than one quality of the good will be offered at the equilibrium.

To illustrate, without any loss of generality, consider the discrete case of only two income levels in each country: consumers with income levels y_H and y_L in the developing country ( $y_H^{}>y_L^{}$ ); and consumers with income levels $y_H^{*}$ and $y_L^{*}$ in the rich country ( $y_H^{*}>y_L^{*}$ ). The competitive firms in the developing country will offer two distinctively different qualities, Q_HC and $Q_{LC}$ , to the rich and poor buyers there, such that

$U_Q(y_H,Q_{HC})=c_Q(Q_{HC}),U_Q(y_L,Q_{LC})=c_Q(Q_{LC})$(5)

Again, by the property of the utility and cost functions, Q_HC  >  Q_LC. Similarly, competitive producers in the rich country will offer two distinctively different qualities, $Q_{HC}^{*}$ and $Q_{LC}^{*}$ , to the rich and poor buyers there, such that

$U_Q(y_H^{*},Q_{HC}^{*})=c_Q(Q_{HC}^{*}),U_Q(y_L^{*},Q_{LC}^{*})=c_Q(Q_{LC}^{*})$(6)

These choices in each country can be illustrated in the same way as in Figure 4 by reinterpreting cross-country income variations as within-country income disparities. That is, we can interpret the steeper (broken) upward-sloping concave curves as representing the iso-net-utility curves for a representative high-income consumer in each country, and the flatter (solid) curves as representing the iso-net-utility curves for a representative low-income consumer in each country.

A few observations are in order. First, high-income buyers in each country will not mimic low-income buyers. For the developing country, for example, high-income buyers will buy the menu $\left\{P_H=c(Q_{HC}),Q_{HC}\right\}$ targeted for them rather than the menu $\left\{P_L=c(Q_{LC}),Q_{LC}\right\}$ targeted for the poorer buyers there because it gives them strictly higher net utility. Note, when more than one menu is offered in the market, each buyer must also decide which menu to buy. The high-income buyers will buy menu {P_H, Q_H} instead of menu {P_L, Q_L} if:

$V(y_H,P_H,Q_H)\ge V(y_H,P_L,Q_L)$(7)

This is called the incentive compatibility (IC) or self-selection (SS) constraint. The tie-breaking rule applied here is that if a buyer is indifferent between the two menus, $V(y_H,P_H,Q_H)=V(y_H,P_L,Q_L)$ , she buys the menu that offers a higher quality. From Figure 4, it can be easily checked that the above IC (or SS) constraint is satisfied for the high-income buyer as a strict inequality for the optimum quality levels Q_HC and Q_LC defined in Equation (5).

Second, if $y_L<y_L^{*}$ and $y_H<y_H^{*}$ , then by Equations (5) and (6), $Q_{LC}<Q_{LC}^{*}$ and $Q_{HC}<Q_{HC}^{*}$ . In such a case, we have quality degradation for all buyers in the low-income developing country that again explains the low-export-quality phenomenon there. Finally, as in the case of identical consumers, in this case of heterogeneous consumers the PP constraint may be relevant for explaining the low-export-quality phenomenon.

3.1.3 Within-Country Income Distribution and Quality Choice: General Equilibrium Analysis

Among the few general equilibrium analyses of the non-homothetic preferences of consumers and the income disparities among them, the home-market effect in Reference Fajgelbaum, Grossman and HelpmanFajgelbaum et al. (2011) explains why richer countries export higher-quality goods. This is similar to the Reference LinderLinder (1961) hypothesis that a country produces goods according to the tastes (or demand) of their local consumers and sells to countries with similar tastes or demand patterns. To explain, they considered two countries, Rich and Poor, which have identical technology and factor endowments but different size and distribution of their labour forces. Each country produces two types of good, a homogeneous good, which serves as the numeraire, and a set of differentiated products. Both types of good are produced only by labour. One unit of labour is required to produce one unit of the homogeneous good. On the other hand, similar to Reference KrugmanKrugman (1979), the differentiated products, which have both variety and quality dimensions, require fixed units of labour per variety as well as per unit of output. Whereas the homogeneous good is produced under competitive conditions, the differentiated goods are produced by monopolistically competitive producers.

Each country has a continuum of individuals endowed with different amounts of labour, and consequently with different wage incomes.Footnote ¹² That is, heterogeneity in endowments generates a distribution of income. Each individual h consumes the homogeneous good and her optimal choice of variety and quality from the set of differentiated products. Individual varieties of the differentiated product reflect attributes of the product other than its quality. Her utility function displays complementarity between the quantity of the homogeneous good and the quality of the differentiated product as well as an idiosyncratic taste component that captures her valuation of the attributes of the differentiated products:

$u_j^h=xQ+{\epsilon }_j^h$(8)

where $u_j^h$ is the utility that she derives by consuming x units of the homogeneous good and one unit of variety j of the differentiated product with quality Q; and ${\epsilon }_j^h$ is her evaluation of the particular attributes of variety j. The complementarity property of this utility function implies a higher MU from a particular quality for those who consume more of the homogeneous good, which in turn implies two things. First, aggregate demand is non-homothetic. Second, since the individual h spends her residual income y^h − p_j on the homogeneous good after buying one unit of her most preferred variety of the differentiated product at the price p_j, so the MU from quality is increasing in her income similar to the specifications of Reference Gabszewicz and ThisseGabszewicz and Thisse (1979) and Reference Shaked and SuttonShaked and Sutton (1982).

A firm that produces a variety j with quality q of the differentiated product, using f_q units of labour to develop the variety and c_q units of labour per unit of output of quality q, earns profit equal to ${\pi }_j=(p_j-c_q)d_j-f_q$ , where d_j is the aggregate demand for variety j. The profit-maximising price set by the firm is then shown to be a mark-up over the (constant) MC, c_q, with the mark-up differing for goods of different quality.

Given this set-up, trade will be governed entirely by demand differences: either differences in the size of labour forces allowing for different income levels or distribution of labour forces generating differences in distribution of incomes. Assuming that the differentiated products are costly to trade, Reference Fajgelbaum, Grossman and HelpmanFajgelbaum et al. (2011) established the following results. First, under autarchy, there is a larger number of firms producing high-quality goods than there are firms producing low-quality products in the larger country. The opposite holds for the smaller country. Second, asymmetry in the relative number of firms producing high-quality goods in the two countries carries over to free trade with incomplete specialisation. Third, similar results hold when the two countries are identical in size but the distribution in the richer country first-order stochastically dominates that in the poorer country. This is because a more dispersed distribution of income in the richer country results in a larger home demand for high-quality goods and a smaller home demand for low-quality goods. More firms then enter the market to produce high-quality goods in the more unequal richer country.

From these results it follows that the richer country will be a net exporter of higher-quality goods and a net importer of lower-quality goods. This, as Reference Fajgelbaum, Grossman and HelpmanFajgelbaum et al. (2011, p. 724) exemplify, explains ‘why Germany traditionally has exported high-quality cars to Korea while importing low-quality cars from there’.

3.2 Supply-Side Explanations: Technology, Factor Endowment and Cost of Quality

Supply-side explanations of the low-export-quality phenomenon in developing countries include market structure (or nature of entry and competition among firms), technology and costs. In Section 3.2.1 we discuss the role of market structure in the context of the partial equilibrium endogenous quality choice models discussed in Section 3.1.1. The contestable nature of markets in the developing countries with hit-and-run entry is another plausible explanation of the low-export-quality phenomenon that we discuss in Section 3.2.2. Technology determines the costs of quality and consequently provision of quality. This aspect of cross-country technology asymmetry as an explanation is discussed in Section 3.2.3. In this context, general equilibrium analyses following Reference Falvey, Kierzkowski and KierzkowskiFalvey and Kierzkowski (1987), Reference Flam and HelpmanFlam and Helpman (1987), Reference Acharyya and JonesAcharyya and Jones (2001) and Reference Ganguly and AcharyyaGanguly and Acharyya (2020, Reference Ganguly and Acharyya2021), among others, allow us to focus on the endogeneity of costs of quality in contrast to exogenously given costs in the partial equilibrium models. Such analyses also enable us to relate the low-export-quality phenomenon to scarcity of domestic factors of production in the developing countries.

3.2.1 Monopoly, Patents and Under-Provision of Quality

Consider the benchmark analysis of Section 3.1.1 where consumers in each country have the same income level (as well as tastes) and thus have the same MWP but have a lower income level if they belong to a developing country than if they belong to a rich country. The same low-export-quality phenomenon derived in such a context arises if domestic monopolies, instead of competitive firms, provide the good under autarchy, which segments the national markets. In fact, the domestic monopolies will offer the same qualities as the competitive firms:

$Q_m=Q_C<Q_C^{*}=Q_m^{*}\forall y<y^{*}$(9)

This follows from the vertically parallel property of the iso-net-utility curves mentioned earlier.Footnote ¹³

But, in each country the domestic monopolist will charge the maximum price that the consumers will be willing to pay for that level of quality, as given by their IR constraint, and thus extract all the surpluses from them by leaving them with zero net utility. In other words, under monopoly provision of the quality-differentiated good and with identical consumers in each country, there will be price distortion only,with no quality distortion. So, up to this point, market structure or monopoly provision of quality does not add any new dimension to the low-export-quality phenomenon.

However, if we consider the case of within-country income disparities and thus heterogeneous consumers in each country, then monopoly provision of quality will have some additional implications. Two results are particularly relevant for our present discussion. First is that a monopolist may not cater to all consumers depending on the within-country distribution of consumers; second is that there will be under-provision of quality by a monopolist when it caters to all with a separating (or different price–quality) menu.

To illustrate, consider again the discrete case of only two types of buyer in each country as specified earlier. Let the number of consumers having these income levels be, respectively, n_H, n_L, n^*_H and n^*_L. As shown in Reference AcharyyaAcharyya (1998, Reference Acharyya2005), the monopolist will serve both buyers in the developing country with a separating menu – two distinctly different price–quality pairs – if the number of poor buyers is sufficiently large in the following sense:

$\frac{n_L}{n_H}>\frac{U_Q(y_H,Q_L)-U_Q(y_L,Q_L)}{U_Q(y_L,Q_L)}$(10)

If the poorer buyers are also served by a menu $\left\{P_L=U(y_L,Q_L),Q_L\right\}$ that extracts all surpluses from them, the monopolist must offer high-income buyers a menu {P_H, Q_H} that satisfies their IC constraint in Equation (7) as a strict equality, which will leave them with some positive surpluses. That is, while the monopolist gains some profit by catering to the low-income buyers, it loses some profit by not being able to extract all surpluses from the high-income buyers as a consequence. Note that this loss arises because the monopolist does not know which buyers are high-income buyers and, even if it had known, it cannot segment the market to prevent arbitrage among low-income and high-income buyers. Thus, the number of low-income buyers must be sufficiently large to compensate for the loss of profit incurred by catering to them. This critical mass of low-income buyers relative to the mass of high-income buyers is given by the right-hand-side value in Equation (10), which indicates the extent to which the MWP of the high-income buyers exceeds the MWP of the low-income buyers in the developing country. If this condition is not satisfied, then the monopolist caters only to the high-income buyers by offering the menu $\left\{P_H=U(y_H,Q_{HM}),Q_{HM}\right\}$ , thereby extracting all surpluses from them, where Q_HM = Q_HC. This is the case of partial market coverage.

In the case of full market coverage, that is, for income distribution satisfying the condition in Equation (10), while the richer buyers will still be offered the same quality as under partial market coverage (and as provided by the competitive firms), the poorer buyers will be offered a lower quality than the competitive firms: Q_LM < Q_LC. This result is known as under-provision of quality by a monopolist, or quality distortion at the bottom. It arises because by degrading quality for poorer buyers, the monopolist can charge the high-income buyers a higher price and thus can extract a larger surplus from them than it could if it had offered the poorer buyers the same quality as the competitive firms. To what extent the quality will be degraded or under-provided, however, will depend on the mass of poor buyers. The smaller they are in number, the larger will be the quality degradation or the extent of under-provision of quality. At the extreme, when they are very small in number in the sense of reverse inequality in Equation (10), they will not be served at all, whereas the high-income buyers will be charged the maximum price that they are willing to pay for the quality Q_HM = Q_HC.

This under-provision result may explain the low-export-quality phenomenon in many ways. One case is where the rich country market is partially covered – the number of poor buyers there is too small to satisfy the condition similar to Equation (10) – and the developing country market is fully covered – too many low-income buyers there relative to high-income buyers. Another case is where the quality-differentiated good is produced by competitive firms in the rich country and by a monopolist in the developing country.Footnote ¹⁴ In this case of different market structures, we will have $Q_{LM}<Q_{LC}^{*}$ and. $Q_{HM}\le Q_{HC}^{*}$ .Footnote ¹⁵

The above results can also explain low quality in the developing country when the quality-differentiated good under consideration is a patented product and is produced and provided by a patent-holder multinational corporation (MNC) in all markets under the patent protection. With no within-country income disparity, the MNC will offer the same pair of menus as the competitive firms (or domestic monopolies under autarchy in the case of a non-patented product) if it can segment the national markets and stop arbitrage across the two markets: a lower quality (with a lower price) to the low-income country and a higher quality with a higher price in the rich country due to income differences. However, if it cannot stop arbitrage by segmenting the markets, then it must ensure that the buyers in the rich country will not buy the menu offered to the buyers in the low-income developing country.Footnote ¹⁶ That is, it must offer menus that are compatible with the SS or IC constraints of the buyers. And by the under-provision of quality result discussed earlier in the case of within-country income disparity, the MNC will further degrade the quality in the low-income developing country.

3.2.3 Technology, Factor Endowment and Domestic Factor Cost

The partial equilibrium endogenous quality choice models also bring out the role of technology and costs in many ways, as discussed in Reference AcharyyaAcharyya (2005). One case is that additional costs for higher quality are larger in developing countries than in the richer, more advanced counties due to poor technology there:

$c(Q)>c^{*}(Q),c_Q>c_Q^{*}>0,c_{QQ}>c_{QQ}^{*}>0\forall Q$(11)

Then, replicating the earlier arguments, it can be seen that even if MWP is the same across all consumers, due to identical per capita income, firms in the technologically backward developing country will offer lower quality than firms in technologically advanced countries. Moreover, as shown in Reference AcharyyaAcharyya (1998), if the superior technology in advanced countries is such that the MC of quality is either invariant with respect to quality or slowly rising in quality in the sense that for the poorest buyer $U_Q({\underset{\_}{y}}^{*},\overline{Q})>c_Q(\overline{Q})$ , then the topmost quality $\overline{Q}$ will be offered by the firms there.

In general equilibrium, costs are endogenous and depend on the prices of the factors required to produce quality-differentiated goods. If the quality-differentiated goods are produced only by domestic factors like capital and/or skilled labour, then both larger units of such factors required per unit of output – for a particular quality as well as across different quality levels – due to poor technology, and the relative scarcity of such factors in the developing countries result in larger marginal costs for higher qualities.Footnote ¹⁹ Technological inferiority and factor scarcity thus force them to specialise and produce a range of low qualities compared to the high-quality products produced by the rich and more advanced countries. For example, Reference Flam and HelpmanFlam and Helpman (1987) offer an explanation for such specialisation patterns in terms of Ricardian comparative cost advantages arising from technological differences.Footnote ²⁰ Reference Murphy and ShleiferMurphy and Shleifer (1997), on the other hand, emphasise the comparative advantage of the richer countries in producing high-quality goods because of the relative abundance of human capital there. Thus, developing countries, being scarce in human capital, produce low-quality goods. A similar factor endowment explanation is offered in Reference Acharyya and JonesAcharyya and Jones (2001) and Reference Ganguly and AcharyyaGanguly and Acharyya (2021), whereas differences in both factor endowment and technology are emphasised in Reference Falvey, Kierzkowski and KierzkowskiFalvey and Kierzkowski (1987).

To illustrate the role of technology, as in Reference Flam and HelpmanFlam and Helpman (1987), consider the same set of goods as defined earlier: a homogeneous good, serving as the numeraire, and a continuum of quality-differentiated goods. A consumer with income level y buys her desired quantity x of the homogeneous good but one unit of the desired quality of the differentiated good that maximises her utility U(x,Q) subject to the budget constraint $y\ge x+p(Q)$ . The utility function is assumed to display the same complementarity property as discussed earlier. All goods are produced only by labour. While the homogeneous good requires one unit of labour per unit of output, the continuum of quality-differentiated goods indexed by the quality levels Q $\in [0,\overline{Q}]$ require a(Q) and a*(Q) units of labour per unit of output of quality Q in the developing country and in the advanced country, respectively. These per unit labour requirements, determined by the technology available to the countries under consideration, increase in the quality level; but the relative labour requirements, a(Q)/a*(Q), decrease in the quality level. The latter assumption implies that the developing country’s comparative advantage diminishes as we move from lower- to higher-quality goods. This production structure resembles the one in the Ricardian continuum of goods model à la Reference Dornbusch, Fischer and SamuelsonDornbusch, Fischer and Samuelson (1977). If the homogeneous good is produced only in the developing country, the wage rate measured in terms of it is equal to one there. Thus, a good of quality Q₀ $\in [0,\overline{Q}]$ is produced in the developing country if $a(Q_0)\le w^{*}{a}^{*}(Q_0)$ and accordingly the (supply) price of the differentiated good with quality Q is given by:

$p(Q)=\min [a(Q),w^{*}{a}^{*}(Q^{*})]$(12)

Let $\tilde{Q}\in [0,\overline{Q}]$ be such that $a(\tilde{Q})=w^{*}{a}^{*}(\tilde{Q})$ . Then, by the assumed chain of comparative advantages, the developing country will be producing and supplying low-quality goods in the range $[0,\tilde{Q}]$ , whereas the advanced country produces high-quality goods in the range $[\tilde{Q},\overline{Q}]$ . Since the assumed chain of comparative advantage underlying this production specialisation along the continuum reflects the inferior technology of the developing country for higher-quality goods, so technological inferiority explains the low-export-quality phenomenon.Footnote ²¹

In Reference Falvey, Kierzkowski and KierzkowskiFalvey and Kierzkowski (1987), a slightly different production structure of the quality-differentiated goods brings out the role of factor endowments along with the technology. They allow for cross-country technology differences in production of the homogeneous good but the same technology for producing the continuum of qualities. Whereas the homogeneous good requires b and b^* units of labour (L) in the two countries, the differentiated good of quality Q requires one unit of labour and Q units of capital (K) in both the countries. The K/L ratio is fixed for any given quality but increases in quality: Higher quality of the good requires more of K per unit of output. Thus, whereas the X sector is Ricardian type, displaying cross-country technology differences, the Z sector producing the quality-differentiated good has a Heckscher–Ohlin production structure with identical technology across countries.

Perfect competition prevails everywhere so that the zero-profit conditions in the developing country, for example, imply:

$1=bw$(13)

$p\left(Q\right)=w+Qr$(14)

Full employment conditions are given by:

$bX+Z=L$(15)

${\displaystyle {\int }_0^{1}QZ(Q)dQ}=K$(16)

where X is the output of homogeneous good, Z(Q) is the output of a differentiated good with quality Q and Z denotes the total output of the continuum of quality-differentiated goods taken together. Similar conditions with relevant variables with asterisks describe the rich country specifications.

On the demand side, each consumer demands her preferred quality, which, given the relative prices, is uniquely determined by her income. A consumer with higher income demands a higher quality. Thus, consumers with different incomes buy different qualities. However, unlike Reference Flam and HelpmanFlam and Helpman (1987) and Reference Fajgelbaum, Grossman and HelpmanFajgelbaum et al. (2011), a consumer buys more than one unit of the differentiated good of her desired quality. The utility of a consumer with income y thus depends on the units of the homogeneous good (x) and on both the units and the quality of the differentiated good, z and Q: U(x, z, Q). Given the prices defined earlier, the desired quality and quantities are decided by her by maximising utility subject to the budget constraint $y=x+p(Q)z=xwz+rQz$ .

Given these specifications, intra-industry trade in quality-differentiated products will be characterised as follows. Both an inferior technology in the homogeneous product and scarcity of capital would mean that the developing country will be exporting low-quality varieties to the rich country and will import higher-quality varieties from there. To explain, note that, for reasons similar to that spelled out earlier, the post-trade price of a particular quality will be the minimum of the unit cost of production in the two countries:

$p(Q)=\min [w+Qr,w^{*}+Q{r}^{*}]\forall Q$(17)

Consider, for example, the case where the rich country has superior technology for producing the homogeneous good, b^* < b, and both countries produce this good under free trade. Thus, the wage rate in the developing country will be lower because of lower marginal productivity than in the advanced country. The wage cost of producing any quality of the differentiated product will thus be smaller in the developing country. If, however, the developing country is capital scarce, then the rate of return to capital will be higher there than in the advanced country, putting it at a capital-cost disadvantage. Since capital intensity rises with the quality of the good, so, beyond a critical level of quality, the capital-cost disadvantage will be more pronounced than the labour-cost advantage of the developing country. That is, the developing country will have comparative cost advantage for all qualities lower than this critical level of quality, $\tilde{Q}$ , and the advanced country will have comparative cost advantage for all qualities lower than this, where:

$w+\tilde{Q}r=w^{*}+\tilde{Q}{r}^{*}$(18)

This pattern of production specialisation is illustrated in Figure 6. In each country the richer consumers will buy the higher-quality varieties produced by the advanced country and the low-income consumers will buy the lower-quality varieties produced by the developing country. The income disparities explain the existence of demand for all qualities in the continuum and hence intra-industry trade in quality-differentiated goods. But, again, the post-trade specialisation in qualities is determined primarily by supply-side determinants such as technology and factor endowment.

Figure 6 Post-trade production specialisation

Reference Acharyya and JonesAcharyya and Jones (2001) and Reference Ganguly and AcharyyaGanguly and Acharyya (2021) differ from Reference Falvey, Kierzkowski and KierzkowskiFalvey and Kierzkowski (1987) – and also from Reference Flam and HelpmanFlam and Helpman (1987) – in two ways. First, they differentiate between the skill levels required to produce the quality-differentiated goods and the two sets of homogeneous goods – traded and non-traded. Second, they study how redistribution of income across different domestic factors of production, resulting from changes in factor endowment, technology or government policy, can affect export quality.

They consider a small, open economy where competitive firms select a quality of the differentiated export good from a set of technologically feasible qualities. The world price of this export good, Z, varies positively with the selected quality that reflects the higher MWP of foreign buyers for a better quality of the (imported) good that they consume: $P_Z^W=P_Z^W(Q)$ , $P_Z^{W^{\prime }}(Q)>0$ , $P_Z^{W^{″}}(Q)>0$. This good is produced by skilled labour (S) and capital in fixed proportion for a particular quality. For a better quality of the good, Reference Acharyya and JonesAcharyya and Jones (2001) assumed a larger per unit requirement of only the capital input:

$a_{KZ}=a_{KZ}(Q),{a^{\prime }}_{KZ}(Q)>0,{a^{″}}_{KZ}(Q)>0$(19)

The economy also produces two homogeneous goods – a composite trade good (T) and a non-traded good (N) – using unskilled labour and the same capital as used by export good Z. This (T,N) nugget displays Heckscher–Ohlin–Samuelson properties.Footnote ²² Suppose that export good Z is not domestically consumedFootnote ²³ and that consumers have homothetic (and identical) tastes for the composite traded good and the non-traded good.

Competitive conditions everywhere mean price equals average cost:

$P_Z^W(Q)=a_{SZ}{w}_S+a_{KZ}(Q)r$(20)

$P_T^W=a_{LT}w+a_{KT}r$(21)

$P_N=a_{LN}w+a_{KN}r$(22)

where, w is the unskilled wage and $a_{ij};i=S,L,K;j=Z,T,N$ , denote the per unit requirement of input-i in production of good-j. While $a_{SZ}$ and a_KZ are technologically fixed, as mentioned earlier, a_ij s in the (T, N) nugget represents the least-cost choices made by the producers that depend only on the wage–rental ratio under the assumption of Constant Returns to Scale (CRS) technology.

For any given rate of return to capital, competitive producers choose the profit-maximising quality $\tilde{Q}$ of Z such that the marginal revenue equals the MC for this quality level:

$P_Z^W{}^{\prime }(\tilde{Q})={a^{\prime }}_{KZ}(\tilde{Q})r$(23)

The price of the non-traded good is determined locally and must be the one for which the domestic market clears. Under homothetic tastes, this means:

$\frac{D_N}{D_T}=f\left(\frac{P_N}{P_T^W}\right)=\frac{X_N}{X_T}$(24)

where X_T and X_N are output levels of the traded and non-traded goods, respectively.

Finally, flexible factor prices and competition ensure that the domestic factor markets clear as well:

$\overline{K}-a_{KZ}(Q)X_Z=\tilde{K}(Q)=a_{KT}{X}_T+a_{KN}{X}_N$(25)

$\overline{S}=a_{SZ}{X}_Z$(26)

$\overline{L}=a_{LT}{X}_T+a_{LN}{X}_N$(27)

Note that by assuming a fixed per unit requirement of skilled labour for all quality levels, Reference Acharyya and JonesAcharyya and Jones (2001) essentially pin down the output of the quality-differentiated export good Z by the availability of skilled labour.

Referring back to the marginal condition for quality choice shown in Equation (23), note that a ceteris paribus fall in the rate of return to capital lowers the MC at any given level of quality and thus induces firms to upgrade quality. This is captured through the downward-sloping Π_Z curve in Figure 7. On the other hand, a ceteris paribus increase in the export quality raises the demand for capital and thus lowers its availability for the (T, N) nugget. Consequently, similar to the output-magnification effect, the output of the composite traded good falls and that of the non-traded good increases under the assumption that the composite traded good is relatively capital intensive. The non-traded market thus clears at a lower price of the non-traded good, which in turn lowers the unskilled wage and raises the rate of return to capital by the price-magnification effect. This relationship between export quality and rate of return to capital is represented by the positively sloped curve labelled M_N.

Figure 7 Equilibrium export quality

The equilibrium quality and the rate of return to capital are thus determined simultaneously corresponding to the point of intersection of these two curves.

The role of factor endowment in explaining the low-export-quality phenomenon is now evident from the previous discussion. A smaller endowment of capital, or alternatively a larger endowment of unskilled labour, means a higher rate of return to capital and correspondingly a higher MC of quality. Competitive firms thus choose a lower quality of the export good Z. That is, relatively unskilled labour-abundant or capital-scarce countries, typically developing countries, will be producers of low-quality export good(s), whereas relatively capital-abundant countries, typically developed countries, will be producers of high-quality export good(s).

In a more generalised case where higher qualities require more of both skilled labour and capital per unit of output, though at different proportions, as considered in Reference Ganguly and AcharyyaGanguly and Acharyya (2020, Reference Ganguly and Acharyya2021), scarcity of skilled labour in developing countries may also explain low quality there. However, the underlying technology for quality upgrading also becomes important. For example, suppose higher-quality varieties require larger units of both capital and skilled labour per unit of output:

$a_{hZ}=a_{hZ}(Q),{a^{\prime }}_{hZ}(Q)>0,{a^{″}}_{hZ}(Q)>0;h=S,K$(28)

where a_hZ denotes the per unit requirement of input-h.

Thus, the relative capital intensity of a higher-quality variety of good Z can increase or decrease according to the nature of the export good and its technological requirement. More precisely, the relative skill intensity $s_Z=a_{SZ}(Q)/a_{KZ}(Q)$ will increase with 1 per cent betterment of quality if the quality elasticity of the per unit input requirement of skilled labour, γ_SZ, is larger than that of capital, γ_KZ:

${\widehat{s}}_Z=\left({\gamma }_{SZ}-{\gamma }_{KZ}\right)\widehat{Q}>0if{\gamma }_{SZ}>{\gamma }_{KZ}$(29)

where ${\gamma }_{hZ}\equiv \frac{Q{a^{\prime }}_{hZ}(Q)}{a_{hZ}(Q)}>0;h=S,K,$ and a hat over a variable denotes its proportional change (e.g. ${\widehat{s}}_Z=\frac{d{s}_Z}{s_Z}$ ).

The marginal condition Equation (23) now changes to:

$P_Z^W{}^{\prime }(\tilde{Q})={a^{\prime }}_{SZ}(\tilde{Q})w_S+{a^{\prime }}_{KZ}(\tilde{Q})r$(23a)

Since, by the zero-profit condition in Equation (22), the skilled wage and the rate of return vary inversely with each other, so now the ${\Pi }_Z$ curve in Figure 7 will be downward sloping only if ${\gamma }_{SZ}<{\gamma }_{KZ}$ , and positively sloped otherwise. If ${\gamma }_{SZ}>{\gamma }_{KZ}$ , that is, higher qualities are relatively skill intensive, then a developing country having a smaller endowment of skilled workers, ceteris paribus, will be producing lower qualities of good Z than countries with a relatively larger number of skilled workers.Footnote ²⁴

The differences in technological requirements of quality upgrading for different types of good also explain the asymmetric quality variations across different goods that were reported in Section 2. Reference Ganguly and AcharyyaGanguly and Acharyya (2021) consider two types of quality-differentiated export good, Z₁ and Z₂, such that ${\gamma }_{KZ}^1>{\gamma }_{SZ}^1$ , ${\gamma }_{KZ}^2<{\gamma }_{SZ}^2$ . That is, whereas higher-quality varieties of Z₁ are relatively capital intensive, higher-quality varieties of Z₂ are relatively skill intensive. Examples of Z₁-type goods are aerospace products, scientific instruments, defence equipment, household and office equipment, electrical appliances, agro-based products and the like. On the other hand, Z₂ represents goods and services such as software, jewellery, diamond cutting and polishing, IT-enabled services, financial services and the like. Goods Z₁ and Z₂ are produced by skilled labour along with different types of capital. While good Z₁ uses the same type of capital as a homogeneous import-competing good Y, good Z₂ uses a specific type of capital.Footnote ²⁵ In such a context, they establish that an increase in the rate of return to capital used by Z₁ and the import-competing good – due to an adverse endowment shock, for example – lowers the quality of good Z₁ and raises the quality of good Z₂.Footnote ²⁶ This result follows from the fact that a change in the rate of return to the commonly used capital changes the rate of return to the other type of capital specific to Z₂ production in the same direction but the skilled wage in the opposite direction. Consequently, at initial quality levels, changes in the relevant factor prices asymmetrically affect the MC of quality upgrading of Z₁ and Z₂. Thus, with technological requirements for quality upgrading varying across different export goods, capital scarcity may not have adverse implications for the quality of all types of export good.

3.3 Concluding Remarks

The central message of the discussion in this section is how fundamentals such as taste diversity, technology and factor endowments can explain both the nature of specialisation by rich and poor countries along the quality spectrum of goods and the observed low-export-quality phenomenon in developing countries. Moreover, under non-homothetic tastes, income disparities both within and across countries – that may itself be a reflection of uneven distribution of factor endowments within a country as well as cross-country differences in such endowments – also cause poorer countries with relatively higher income inequality to emerge as producers of low-quality goods.

Experience goods with unobservable quality and associated information externality, on the other hand, pose an altogether different kind of problem. Foreign buyers rely on firms’ reputations and perceptions regarding the country of origin of any imported good that they consume. But the contestable nature of markets in developing countries often creates disincentives for firms there to invest in building and maintaining reputation. Poor country-of-origin perceptions for manufactured goods produced in developing countries also discourage firms there from producing anything of a higher quality than the industry average.

4.1 Trade Liberalisation Policies

Tariff liberalisation policies adopted by developed as well as developing countries, ever since their accession to the WTO, are primarily of two types: (1) reduction of tariffs on final import-competing goods and (2) reduction of tariffs on imported inputs. While the significant impact of such tariff cuts on the volume of trade is already recorded in the existing literature, in Sections 4.1.1 and 4.1.2 we analyse their implications for export quality upgrading and consequent gains from trade at the extensive margin.

4.1.1 Tariff Reduction on Final Import Competing Good

It is well known from the celebrated Stolper–Samuelson theorem and its generalisation as the price-magnification effect à la Reference JonesJones (1965) that reduction of tariff on imported final consumption goods will change domestic factor prices asymmetrically, and accordingly will have income redistribution effects. Given the empirical evidence mentioned in Section 3 regarding higher-quality varieties being dependent on more intensive use of domestic factors such as skilled labour and capital, this would imply two things. First, if relative skill intensity increases with the quality level in the sense defined in Section 3 (see Equation (29)), then a tariff reduction may downgrade export quality if it raises the skilled wage relative to the rate of return to capital. In such a case, tariff reduction fails as an export-quality-promoting strategy. Second, if different goods and services require different technologically fixed relative skill intensities in the upgrading of their respective qualities, then tariff reduction on the imported final goods may lead to asymmetric quality variations across such goods and services.

These results were formally established in Reference Ganguly and AcharyyaGanguly and Acharyya (2021). Recall the analytical structure outlined in Section 3. The economy produces two quality-differentiated manufactured goods (or services), Z₁ and Z₂, which differ from each other in two ways. First, these goods use different types of capital, K₁ and K₂, respectively, along with skilled labour in production. Second, whereas higher-quality varieties of Z₁ are relatively capital intensive, higher-quality varieties of Z₂ are relatively skill intensive ( ${\gamma }_{KZ}^1>{\gamma }_{SZ}^1$ ,   ${\gamma }_{KZ}^2<{\gamma }_{SZ}^2$ ). The economy also produces a homogeneous agricultural export good X that is produced by unskilled labour and land, and a homogeneous import-competing good Y produced by unskilled labour and K₁-type capital (which is also used by Z₁). Thus, K₂-type capital and land are specific to the Z₂ and X sectors, respectively. Moreover, in keeping with the observation that labour markets are segmented in most developing countries, Reference Ganguly and AcharyyaGanguly and Acharyya (2021) assume that while the import-competing good is produced in the formal sector where unskilled workers are paid a (fixed) minimum wage ( $\overline{w}$ ), the agricultural export good is produced in the informal sector where unskilled workers are paid a competitive, market-determined wage.Footnote ²⁷ Despite the minimum wage paid in the formal sector, all unskilled workers are fully employed since those who cannot get employed in the formal import-competing sector are absorbed in the informal agricultural sector through the necessary fall in unskilled wages there.

In such a set-up, the following zero-profit and marginal conditions help elucidate the relationship between factor prices and export quality:

$P_{Zm}^W(Q_m)=a_{KZ}^m(Q_m)r_m+a_{SZ}^m(Q_m)w_S,m=1,2$(30)

$P_X^W=a_{LX}w+a_{TX}R$(31)

$P_Y=(1+t)P_Y^W=a_{LY}\overline{w}+a_{KY}{r}_1$(32)

$P_{Zm}^W{}^{\prime }(Q_m^0)=a_{KZ}^m{}^{\prime }(Q_m^0)r_m+a_{SZ}^m{}^{\prime }(Q_m^0)w_S,m=1,2$(33)

where R is the rate of return to land, t < 1 is the rate of ad valorem tariff, and other notations are as specified earlier.

Three observations are in order. First, the quality choices depend only on the two policy parameters – tariff and minimum wage – but not on endowments of two types of capital and skilled labour. Given the fixed minimum wage in the Y sector, the rate of return to K₁-type capital is tied to the tariff-inclusive price of imports P_Y. On the other hand, as evident from Equation (30), given this rate of return to capital, first the skilled wage is determined by the world price of Z₁, and then, given the skilled wage, the rate of return to K₂-type capital is determined by the world price of Z₂. That is, given $\overline{w}$ , the rates of return to the two types of capital and the skilled wage are determined solely by the world prices for quality-differentiated goods and the tariff-inclusive price of the import good. Since only these factor prices are relevant for quality choices, the quality levels are independent of endowments. Second, quality variations affect unskilled wage and the return to land by changing the availability of K₁-type capital for the (X, Y) nugget and the output composition there. Although this has some far-reaching implications for income redistribution (or wage inequality), as we will discuss in Section 5, it is itself inconsequential for quality choices. Third, as mentioned in Section 3, whereas r₁ and r₂ move in the same direction in the event of any policy shock, the skilled wage moves in the opposite direction. This is evident from a closer look at Equation (30). Thus, in both quality-differentiated sectors, the skilled labour to capital cost moves uniformly, which sets the stage for asymmetric quality variations due to different technological requirements for improving Q₁ and Q₂.

Now, consider a ceteris paribus reduction in the tariff rate. To begin with, in the formal sector, import-competing production contracts due to increased foreign competition, which lowers demand for both unskilled labour and capital, causing the informal unskilled wage and the rate of return to K₁-type capital (r₁) to fall. At the initial Q₁, the consequent lower capital cost encourages producers of Z₁ to expand production, which raises both the demand for skilled workers and their wage. Consequent upon this fall in r₁ and rise in w_S, the MC of quality of Z₁ falls because quality upgrading of Z₁ requires more intensive use of capital than skilled labour ( ${\gamma }_{KZ}^1>{\gamma }_{SZ}^1$ ). Producers of Z₁ thus upgrade their quality. On the other hand, at the initial Q₂, the rise in the skilled wage lowers the production of Z₂, which in turn lowers the rate of return to the K₂-type capital specific to this sector. But, since ${\gamma }_{KZ}^2<{\gamma }_{SZ}^2$ , the MC of quality rises and producers lower the quality Q₂.

Therefore, reduction of a tariff on imported final consumption goods causes the quality of Z₁ and Z₂ to change asymmetrically and in a contrasting way due to their different technological requirements for domestic inputs for quality upgrading. This provides a plausible explanation of the asymmetric quality variations across different export products as documented in Section 2.

4.1.2 Tariff Reduction on Imported Input

The dependence of production of final traded goods on imported intermediates can be typically observed in many developing countries. Examples of such import inputs are petroleum, oil and lubricants, gold and precious metals for jewellery setting, electrical and electronic components, and CC (chemical elements and compounds). This dependence is partly due to either unavailability of such inputs or lack of good-quality indigenous inputs. High tariffs on imports of inputs and intermediates, put in place in many instances to both drive revenue and protect domestic industries, have compelled domestic producers of final goods to use indigenous inputs and/or imported inputs less intensively, which has adversely affected the quality of such goods. In such a context, it is likely that input-tariff reduction would help domestic producers upgrade the quality of the goods they produce and export, through more efficient use of better-quality imported inputs.

Similarly, since product quality is integral to global value chains (GVCs), then facilitating participation in those through reduction of trade costs such as tariffs and non-tariff barriers (NTBs) may provide an incentive for poorer countries to move up the quality ladder. However, as the GVC Development Report (2017) observes, participation of developing countries in GVCs has been rather limited because only a few have sufficient geographical proximity to the three main interconnected hubs, around the United States, Germany in Europe, and China, Japan and South Korea in Asia. Labour productivity is another reason for the limited participation in GVCs. Thus, quality improvement driven by GVCs may be limited to only a few developing countries.

A sizeable recent empirical literature emphasises the favourable impact of input-trade liberalisation on incentivising quality upgrading of exports. For example, Reference Bas and Strauss-KhanBas and Strauss-Khan (2013) find that Chinese firms sourcing high-quality inputs from developed countries experience a rise in export prices and upgraded quality of the products they are exporting to the high-income countries. Further, a within-firm quality upgrading of imported inputs after input-trade liberalisation is observed by Reference Bas and Strauss-KhanBas and Strauss-Khan (2013) as the firms imported not only larger varieties of inputs but also higher-priced inputs. Reference Fan, Li and YeapleFan et al. (2018) also observe a causal effect of input-trade liberalisation on quality upgrading in the context of China’s accession to the WTO in 2001.

Of course, if the quality and the intensity of use of imported input or intermediate goods were the only key factors for quality upgrading, then input-trade liberalisation incentivising exporters to upgrade the quality of their export goods would be a foregone conclusion. But, once again, the trends in asymmetric quality variations across different product groups discussed in Section 2 speak differently. That reduction of input tariffs may not necessarily improve export quality can be demonstrated in two contexts. First, imported inputs are used in the production of export goods other than quality-differentiated export goods; second, imported inputs are used in the production of quality-differentiated export goods along with skilled labour and capital. To demonstrate, recall the analytical structure of Reference Acharyya and JonesAcharyya and Jones (2001) spelt out in Section 3, with the generalised quality-upgrading technology specified in Equation (28) and the corresponding marginal condition in Equation (23a). As a simple extension, suppose the composite traded good uses an imported input (I) in a fixed quantity per unit of output, a_IT. This captures the import dependent production in many developing countries. The import of this input at the given world price P^W_I is, however, subject to an advalorem tariff at the rate $\tau <1$ . Thus, the zero-profit condition in Equation (21) in Section 3 should now be rewritten as

$P_T^W=a_{LT}w+a_{KT}r+a_{IT}(1+\tau )P_I^W$(34)

Henceforth, we will refer to this analytical structure as the extended Acharyya–Jones (A–J) framework. For any given input tariff, the export quality and the rate of return to capital are simultaneously determined as before, as re-illustrated in Figure 8. Recall that, for reasons spelt out earlier, the profit-maximising quality choice varies inversely with the rate of return to capital for ${\gamma }_{SZ}<{\gamma }_{KZ}$ (as in Figure 8 panel (a)) but positively for ${\gamma }_{SZ}>{\gamma }_{KZ}$ (as in Figure 8 panel (b)). Accordingly, the ${\mathrm{\Pi }}_Z$ curve is downward sloping in the left-hand-side panel and upward sloping in the right-hand-side panel. The other relationship between quality and rate of return to capital, which was depicted through the M_N curve, will also have two different slopes under the generalised quality-upgrading technology. Note that now the output of the quality-differentiated export good Z is no longer pinned down by the availability of skilled labour. So full employment of skilled labour (given by Equation (26)) will become:

$\overline{S}=a_{SZ}(Q)X_Z$(35)

Unlike what is represented in Equation (23) earlier, now a ceteris paribus increase in the quality of Z will require not only additional capital but also additional skilled labour, for which output of Z must fall to make such additional skilled workers available for quality upgrading. Consequently, capital demand rises by the extent ${\widehat{a}}_{KZ}={\gamma }_{KZ}\widehat{Q}$ and falls due to this scale contraction by the extent $\widehat{Z}=-{\gamma }_{SZ}\widehat{Q}$ at the margin. Hence, overall, the capital requirement in the production of good Z will rise following quality upgrading if ${\gamma }_{SZ}<{\gamma }_{KZ}$ , and will fall otherwise. In the latter case, ${\gamma }_{SZ}>{\gamma }_{KZ}$ , for example, capital released by the Z sector moves to the (T, N) nugget. Assuming that the composite traded good is relatively capital intensive compared to the non-traded good, this larger availability of capital causes output of the composite traded good to rise and that of the non-traded good to fall by the standard output-magnification effect. Consequent excess demand for the non-traded good raises its price. With P^W_T and P^W_Iunchanged, this rise in P_N will then raise the unskilled wage and lower the rate of return to capital by the standard price-magnification effect.Footnote ²⁸ So a rise in the quality of Z will cause the rate of return to capital to fall when ${\gamma }_{SZ}>{\gamma }_{KZ}$ , for any given rates of input tariff. Similarly, when a higher-quality variety of Z is relatively more capital intensive ( ${\gamma }_{SZ}<{\gamma }_{KZ}$ ), a rise in export quality will raise the rate of return to capital. So the M_N curve is positively sloped if ${\gamma }_{SZ}<{\gamma }_{KZ}$ and negatively sloped if ${\gamma }_{SZ}>{\gamma }_{KZ}$ .

Figure 8 Equilibrium export quality

The equilibrium quality Q₀ of the export good/service Z and the rate of return to capital r₀ are thus determined simultaneously at the intersection of the M_N and ${\mathrm{\Pi }}_Z$ curves for any given rates of input tariff. Now suppose that the tariff on the imported input I is lowered, i.e. $\widehat{\tau }<0$ . This raises the effective price received by the producers of the composite traded good, or the effective marginal revenue earned per unit: ( $P_T^W-a_{IT}(1+\tau )P_I^W$ ). At the initial level of export quality and corresponding P_N, the higher effective price of the composite traded good raises the rate of return to capital (and lowers the unskilled wage) through the price-magnification effect, given the factor intensity assumption that the composite traded good is relatively capital intensive. The M_N curve in both panels in Figure 8 shifts up, causing downgrading of quality if ${\gamma }_{KZ}>{\gamma }_{SZ}$ and upgrading if ${\gamma }_{KZ}<{\gamma }_{SZ}$ .

The case of an imported input being used in the production of a skill-based quality-differentiated good and the impact of a reduction in tariff on this input can similarly be analysed in the extended A–J framework. For example, as shown by Reference Ganguly, Acharyya, Bagli, Chakrabarti and GuhaGanguly and Acharyya (2022a), lowering the tariff on the input used in Z good will improve the quality of Z only if a higher-quality variety of it requires additional imported input to a larger extent than additional skilled labour. Reducing the input tariff lowers the MC of quality for any given skilled wage but, at the same time, raises the MC of quality upgrading by raising the skilled wage, since it augments production of skill-based export goods. Hence, overall, the MC of quality will fall, and quality will be upgraded if relatively more imported input is required for upgrade quality.

In this context, two recent empirical studies deserve attention. First, Reference Fieler, Eslava and XuFieler, Eslava and Xu (2018) develop a blended quantitative model of input-trade liberalisation where higher output quality requires both intermediate inputs to be of high quality. If foreign inputs are superior in quality to domestic inputs, then input-trade liberalisation allows monopolistically competitive heterogeneous firms access to higher-quality inputs and endogenously to choose output quality as a function of their productivity, their use of skilled labour and inputs of different quality. Second, Reference Bas and PaunovBas and Paunov (2021) show that Ecuador’s firms upgraded both the quality of their imported inputs and their skill intensity following input-tariff cuts there during 1997–2007. Using instrumental variables estimation, they find that the more skill-intensive firms that import better-quality inputs are also the ones that upgrade output quality.

To summarise, whether tariff reductions upgrade or downgrade quality depends on the nature of the export good, that is, on the relative skill intensity of its higher-quality varieties. Moreover, when quality-differentiated export goods of a developing country differ from each other in terms of the relative skill intensity of their respective higher-quality varieties, tariff reduction on imported consumption goods does not uniformly affect the quality of such goods. Reduction of input tariffs also affects the quality of different types of export good asymmetrically, depending on the intensity of the imported inputs relative to the domestic inputs. All these provide plausible explanations of the asymmetric quality variations across different export products that we observed earlier for quite a few developing countries.

4.2 Exchange Rate Policies

Another standard trade policy usually adopted by developing countries with the aim of making exports to the rich world cheaper is exchange rate policy. This takes the form of currency devaluation in countries under a pegged regime – for example, most of the oil-exporting countries in the Middle East and Asia – and policy interventions to moderate exchange rate appreciations that erode the price competitiveness of exports in countries that have adopted a managed float. The links between exchange rate fluctuations and the corresponding adjustments in export prices and export volumes have been well explored (Reference McKenzieMcKenzie, 1999; Reference Martín and RodríguezMartín and Rodríguez, 2004; Reference Gopinath and RigobonGopinath and Rigobon, 2008; Reference Thorbecke and SmithThorbecke and Simith, 2010; Reference Berman, Martin and MayerBerman, Martin and Mayer, 2012; Reference Li, Ma and XuLi, Ma and Xu, 2015). However, the implications for quality choice and the consequent gains from trade in a world that values higher quality over cheaper varieties of goods and services remain less investigated.

Among the few existing studies, Reference YuYu (2013) theoretically explores how exchange rate pass-through depends on firm heterogeneity in productivity and product differentiation in quality. Extending the model devised byReference AntoniadesAntoniades (2008), Reference YuYu (2013) considers the quantity-dependent component of quality-upgrading cost in addition to the quantity-invariant R&D cost of quality upgrading. This allows an exporting firm to choose different quality levels for different markets in order to absorb the exchange rate changes, in addition to adjusting its mark-ups given the linear demand structure. While this analysis sheds some light on the link between exchange rate fluctuations and quality of export goods, it does not decompose domestic factor costs into skill and capital components that may or may not move in tandem as a consequence of inter-sectoral reallocations of these resources following exchange rate shocks. Hu et al. (2014), on the other hand, highlight the impact of exchange rates on firm-level imported intermediates and argue that a decrease in firm-level import exchange rate (i.e. an appreciation in the domestic currency) enables firms to buy higher-quality intermediates, which they could not have afforded before. Hu et al. (2014) verify their theoretical model’s predictions using a rich and unique database of Chinese firms’ trade data. Reference Chen and JuvenalChen and Juvenal (2014) show that the elasticity of demand perceived by exporters decreases with both real depreciation and quality. In response to real depreciation, firms significantly increase more of their mark-ups and less of their export volumes for higher-quality products. Given that higher-income countries have a stronger preference for higher-quality goods, this heterogeneous response of prices and quantities to exchange rate changes is predicted to be stronger for exports to high-income destination countries. The predictions are empirically tested using disaggregated Argentinean firm-level wine export values and volumes between 2002 and 2009, with experts’ wine ratings as a measure of quality. More recently, Reference Chen, Lu and TianChen, Lu and Tian (2021) establish that credit-constrained firms may upgrade their export product quality in response to foreign currency appreciation. However, they consider labour to be the only factor of production and do not distinguish between different types of export good in terms of their factor (domestic as well as imported) intensities and product quality. In contrast, the theoretical analysis of Reference Ganguly and AcharyyaGanguly and Acharyya (2022b) shows that, though currency devaluation may raise revenue from exports and thus incentivise quality choice, quality upgrading is still not a foregone conclusion because it also raises the cost of the imported and domestic inputs necessary for quality upgrading. Their result can be illustrated in the extended A–J framework with a rigid unskilled money wage throughout the economy. Assuming e to be the officially pegged exchange rate – units of domestic currency per unit of foreign currency (say, US dollar) – the price-average cost conditions in the (T, N) nugget should now be rewritten as:

$P_T^{}=e{P}_T^W=a_{LT}\overline{w}+a_{KT}r+a_{IT}(1+\tau )e{P}_I^W$(36)

$P_N=a_{LN}\overline{w}+a_{KN}r$(37)

Due to money wage rigidity, the rate of return to capital is solely and uniquely determined by the nominal exchange rate, given the state of technology, the imported input tariff rate and the world prices of T and I. This, in turn, determines the skilled wage for any given level of export quality from the zero-profit condition (refer back to Equation (26)). So the factor prices are delinked and independent from quality and output changes, and are determined solely by policy for any given state of technology, price of the composite traded good and price of the imported input. Note that the price of non-traded goods is now cost determined.

In this set-up, a devaluation of the domestic currency generates two effects. First, marginal revenue earned from per unit production of good Z rises by the exact rate of currency devaluation at the initial choice of quality and the corresponding world price in foreign currency P^W_Z. At the initial rate of return to capital, this would cause the skilled wage to rise more than proportionately by the magnitude $\frac{\widehat{e}}{{\theta }_{SZ}}$ .Footnote ²⁹ Second, devaluation also raises the effective revenue $(e{P}_T^W-a_{IT}(1+\tau )e{P}_I^W)$ earned by producers of the composite traded good by $(1-{\theta }_{IT})\widehat{e}$ . This will induce producers to raise the level of output of this good, which in turn will entail an increased demand for both unskilled labour and capital. Unskilled labour can be drawn from the existing pool of unemployed workers at the fixed money wage. But, capital being fully employed, the increased demand for capital causes its rate of return to rise by a magnitude even greater than the rate of devaluation, $\frac{(1-{\theta }_{IT})\widehat{e}}{{\theta }_{KT}}$ . For any given domestic-currency price of Z, this in turn lowers the skilled wage by the magnitude $\frac{(1-{\theta }_{IT}){\theta }_{KZ}\widehat{e}}{{\theta }_{KT}}$ . Thus, the final change in the skilled wage is ambiguous:

${\widehat{w}}_S=\left[\frac{1}{{\theta }_{SZ}}-\frac{(1-{\theta }_{IT}){\theta }_{KZ}}{{\theta }_{SZ}{\theta }_{KT}}\right]\widehat{e}=\frac{{\theta }_{KT}-(1-{\theta }_{IT}){\theta }_{KZ}}{{\theta }_{SZ}{\theta }_{KT}}\widehat{e}$(38)

However, even when the skilled wage increases, it increases less than proportionate to the rate of devaluation (i.e. ${\widehat{w}}_S-\widehat{e}<0$ ). Thus, $\widehat{r}>\widehat{e}>{\widehat{w}}_S$ .

Now, for change in quality, what matters is whether the MC of quality rises more or less than proportionately to the marginal revenue (which is equal to the rate of devaluation) at the initial level of quality. Again, this depends on the relative skill intensity of the higher quality of export good Z. As can be verified algebraically from the marginal condition, quality is upgraded if ${\gamma }_{SZ}>{\gamma }_{KZ}$ , and downgraded if ${\gamma }_{SZ}<{\gamma }_{KZ}$ . Note that devaluation raises export quality under the same conditions as reduction of tariff on imported inputs used by the composite traded good raises export quality. This is not surprising because both policies expand the production of composite traded goods: while input tariff reduction raises the effective price received by producers, devaluation raises the price received by them in the domestic currency. In either case, increased production raises the demand for both unskilled labour and capital. At the initial level of quality of the export good Z, the excess demand for capital raises its rate of return. Consequent higher capital cost induces producers of Z to lower their production levels and consequently the demand for skilled workers. The skilled wage thus falls. Therefore, the MC of quality rises – and correspondingly the quality is downgraded – if higher qualities of good Z are relatively capital intensive. Devaluation also raises the price (and the marginal revenue) received by producers of Z in the domestic currency, but it is not a large enough increase to outweigh the increase in the MC and thus induce them to upgrade quality.

Again, for an economy that has a heterogeneous export basket, this result indicates that devaluation will have asymmetric effects on the quality of such products. To illustrate, once again consider the analytical structure of Reference Ganguly and AcharyyaGanguly and Acharyya (2021) with two types of quality-differentiated export good, as discussed in Section 4.1.1. For export good Z₁, for which ${\gamma }_{KZ}^1>{\gamma }_{SZ}^1$ and which uses the same capital as an import-competing good, devaluation will unambiguously lower its export quality. Note that, without any imported input, and with the import-competing good produced by K₁-type capital and unskilled labour, devaluation raises the rate of return to capital by $\frac{\widehat{e}}{{\theta }_{KY}}$ and, following a similar logic to that spelt out earlier for the extended A–J framework, changes the skilled wage by ${\widehat{w}}_S=\frac{{\theta }_{KY}-{\theta }_{KZ}^1}{{\theta }_{SZ}^1{\theta }_{KY}}\widehat{e}$ . Again, even if the skilled wage rises, it will rise less than proportionate to the rate of devaluation such that ${\widehat{r}}_1>\widehat{e}>{\widehat{w}}_S$ . Thus, given ${\gamma }_{KZ}^1>{\gamma }_{SZ}^1$ , Q₁ will be downgraded. For the other good, Z₂, for which ${\gamma }_{KZ}^2<{\gamma }_{SZ}^2$ and which is produced by the specific K₂-type capital along with skilled labour, devaluation will improve its quality since ${\widehat{r}}_2>\widehat{e}>{\widehat{w}}_S$ when the skilled wage rises.

These asymmetric effects of devaluation, however, are in sharp contrast with the asymmetric effects of tariff reduction on the import of good Y discussed in Section 4.1.1. The contrasting effect arises because devaluation, unlike tariff reduction, raises the demand for capital and accordingly its rate of return, and correspondingly lowers the skilled wage. The MC of quality thus rises under devaluation for those export goods for which higher quality is relatively more capital intensive, thereby forcing producers of such goods to downgrade quality. These contrasting effects of devaluation and tariff reduction bring out the importance of designing a quality-promoting policy that is specifically tailored to the target group of export products.

4.3 Cross-Country Factor Movements

During the last two decades there has been a surge in emigration of both skilled and unskilled workers, mainly from developing to developed countries. According to the Migration Data Portal, the top three origin countries are India, Mexico and China; up until 2019, these sent out, respectively, 17.5 million, 11.8 million and 10.7 million workers. Despite opening up a source of national income in the form of remittances from these emigrants, such factor outflow directly lowers the productive capacity of the origin economy. Several studies discuss different channels through which emigration of workers may promote exports at the intensive margin. While Reference Rauch and TrindadeRauch and Trindade (2002), Reference Felbermayr and ToubalFelbermayr and Toubal (2012), Reference Genc, Gheasi, Nijkamp and PootGenc et al. (2011) and Reference Ehrhart, Le Goff, Rocher and SinghEhrhart et al. (2014) give a demand-side explanation through networking, Reference Rauch and CasellaRauch and Casella (2003) present the home preference effect. In the context of migration affecting quality of exports (gains at the extensive margin), this networking effect may be relevant in cases such as Reference AkerlofAkerlof’s (1970) lemons problem under asymmetric information of foreign buyers. However, for search goods, i.e. goods with observable quality, the cost effect of emigration might seem more relevant than the demand effect through networking.

It may be obvious that emigration of skilled labour lowers export quality in the native or origin country. A ceteris paribus outflow of skilled labour from developing countries due to higher skilled wages abroad (the pull factor of emigration) pushes up the skilled wage by generating a scarcity. If export quality upgrading requires more intensive use of skilled labour, this will increase the MC of raising quality and thus downgrade export quality. Lower quality goods will lower the MWP of buyers in export destinations, thus further sinking the export prospects of the origin country. This provides an additional dimension to the gains and losses from emigration of skilled workers discussed in the existing literature on brain drain and brain gain. A wide range of arguments have been put forward in such literature from both the origin- and the host-country perspectives (see Reference Commander, Kangasniemi, Winters, Baldwin and WintersCommander, Kangasniemi and Winters, 2004, for an earlier survey). Among the benefits that developed host economies get from immigration of skilled workers, easing out of skill shortages resulting from rapid skill-biased technical change is an important one. For developing origin countries, on the other hand, the literature puts forward two important gains. First, emigration of skilled labour may motivate others to acquire more education and technical skills. As long as only a fraction of them emigrate, there may thus be larger human capital available to the origin country, which in turn may promote its growth. Second, emigrants may contribute to investment, capital formation and growth in their native developing countries through networks and remittances (Reference Wei and BalasubramanyamWei and Balasubramanyam, 2006; Reference Wei, Liu, Lu and YangWei et al., 2017). However, while these benefits of emigration of skilled workers may be realised in the long run, in the short run further skill shortages in already skilled-labour-scarce developing countries may cause exporters to downgrade the quality of their products, thereby retarding export growth.

In contrast, emigration of unskilled labour may seem to be not so worrying since it is not directly used in the quality upgrading of skill-based export goods. But trade literature (Reference Jones, Marjit, Neufeind and RiezmanJones and Marjit, 1992; Reference Marjit and BeladiMarjit and Beladi, 1999; Reference Acharyya, Beladi and KarAcharyya, Beladi and Kar, 2019) often talks about complementarity between wages of different skill types. This implies that emigration of unskilled labour can have adverse implications for export quality upgrading and thus for export prospects through the consequent increase in the skilled wage (Reference Ganguly and AcharyyaGanguly and Acharyya, 2020).

Consider a variant of the analytical structure discussed in Section 4.1.1, with two changes. First, only the Z₁-type quality-differentiated export good is produced along with homogeneous export good X and import-competing good Y. This is to simplify the analysis. Second, in both X and Y sectors, the unskilled wages are flexible. With these simplifications, suppose α and β are the proportions of skilled and unskilled workers, respectively, who emigrate abroad. The full employment conditions can thus be written as:

$\left(1-\alpha \right)\overline{S}=a_{SZ}(Q)Z$(39)

$(1-\beta )\overline{L}=a_{LX}X+a_{LY}Y$(40)

$\overline{T}=a_{TX}X$(41)

$\overline{K}=a_{KZ}(Q)Z+a_{KY}Y$(42)

The property of this structure that deserves attention is that at any given level of quality, the skilled and unskilled wages are complementary to each other, that is, move in the same direction, whereas the rate of return to capital (and also land) changes in the opposite direction. It follows from here that the unskilled wage and the choice of export quality are interdependent, causing each other. Along with the complementarity between skilled and unskilled wages, we can derive a causality from unskilled wage to export quality as represented by the QQ curve in Figure 9, the nature of which depends again on whether ${\gamma }_{SZ}>{\gamma }_{KZ}$ or ${\gamma }_{SZ}>{\gamma }_{KZ}$ . On the other hand, change in export quality causes a change in the unskilled wage through the change in capital availability for production of the import-competing good and consequent change in the composition of output levels in the (X, Y) nugget. Again, the nature of this causal relationship, represented by the ww curve, depends on ${\gamma }_{SZ}$ and ${\gamma }_{KZ}$ , but in exactly the opposite way: positively sloped if ${\gamma }_{SZ}>{\gamma }_{KZ}$ , and negatively sloped otherwise.

Now, an exogenous increase in the emigration rate of unskilled workers raises the unskilled wage due to a consequent fall in unskilled workers’ availability for domestic production of homogeneous traded goods, X and Y. By complementarity of wages, this raises the skilled wage and lowers the rates of return to capital and land. Thus, emigration of unskilled workers raises the quality of export good Z if its higher quality is relatively capital intensive, ${\gamma }_{KZ}>{\gamma }_{KZ}$ , and lowers export quality otherwise, as is evident from the shifts of the ww curve in Figure 9.

Figure 9 Emigration, export quality and unskilled wage

If we allow for endogeneity of emigration with the rate of emigration responding to both source-country factors (push factors) and host- or destination-country factors (pull factors), then quality variations due to any policy or exogenous shock will cause the rate of emigration to change as well. This is evident from the following emigration equilibrium condition:

$w(\beta ,Q)=w^{*}$(43)

Note that workers decide to emigrate by equating the gain from migration to its opportunity cost, which is the domestic wages forgone.

Given this interdependence between export quality and rate of emigration, an exogenous shock that triggers larger emigration of unskilled workers at the initial export-quality level will cause quality upgrading by a larger extent and will induce subsequent multiplier expansion in the emigration rate if ${\gamma }_{KZ}>{\gamma }_{SZ}$ .Footnote ³⁰

To summarise, although unskilled workers are not directly used in the production of quality-differentiated export goods, emigration of such workers may downgrade the quality of export goods. This happens because of the complementarity between emigration of skilled and unskilled workers. Thus, developing countries such as China and India, in particular, from which a significant number of unskilled (as well as skilled) workers emigrate to developed countries, may experience downgrading of the quality of certain of their export good types and, consequently, a dip in earnings from such exports. Moreover, since quality changes also affect the rate of emigration of unskilled workers, an initial shock that triggers emigration of unskilled workers may have a magnified quality-downgrading effect.

4.4 Direct Export Quality Promotion Subsidies

The discussions in the preceding sections have shown that neither export-promotion policies such as tariff cuts nor exchange rate interventions and increased cross-country factor flows have an across-the-board favourable effect on choice of export quality. That is, whether they induce quality upgrading or downgrading depends on the type of quality-differentiated export product, as the different types vary in terms of the domestic (skill and/or capital) and imported inputs needed to produce their higher-quality varieties. In such situations, targeted subsidies, such as production subsidies linked to quality level and/or input subsidies provided to producers of adversely hit export goods, can be effective solutions. In Sections 4.4.1 and 4.4.2 we discuss such concurrent policies.

4.4.1 Quality–Content Production Subsidy

As discussed in Section 4.1.1, when there are two types of quality-differentiated export good, Z₁ and Z₂, tariff reduction on the final import good worsens the quality of that good (Z₂) whose higher qualities are relatively more skill intensive. Suppose that a per unit production subsidy, $\psi (Q_2)$ , the rate of which increases with Q₂, is provided to the producers of good Z₂, such as:Footnote ³¹

$\psi (Q_2)=\frac{1}{2}b{Q}_2^2\forall Q_2\in \left[0,1\right]$(44)

So the zero-profit condition for Z₂ can be rewritten as follows:

$P_{Z2}^W(Q_2)+\frac{1}{2}b{Q}_2^2=a_{KZ}^2(Q_2)r_2+a_{SZ}^2(Q_2)w_S^{}$(45)

The subsidy will raise the effective marginal revenue for quality, but still may not incentivise quality upgrading as the MC of raising quality will rise as well. To see how, recall that while the two goods Z₁ and Z₂ use two different types of capital, they share the same skilled labour. As the subsidised producers intend to raise quality, they will require more units of both skill and K₂-type capital. While skilled labour can be drawn from the other sector, Z₁, at the given skilled wage, higher demand for sector-specific K₂-type capital will raise its rate of return r₂. So, other than the rise in effective marginal revenue, the capital cost of raising quality Q₂ also rises at the margin, rendering the effect of a quality–content production subsidy uncertain. As explained in Reference Ganguly and AcharyyaGanguly and Acharyya (2021), at the lower final import tariff rate following a tariff liberalisation policy, the quality of Z₂ will be upgraded as producers receive a quality–content production subsidy at the rate b, if ${\gamma }_{KZ}^2<2$ . To explain briefly, a 1 per cent increase in the rate of subsidy raises the effective marginal revenue from quality upgrading by 1 per cent. On the other hand, since a 1 per cent increase in the rate of return to capital r₂ raises the MC by ${\theta }_{KZ}^2{\gamma }_{KZ}^2$ , by the quadratic form of the subsidy function, a 1 per cent rise in the rate of production subsidy raises the MC of quality upgrading by $\frac{{\gamma }_{KZ}^2}{2}$ per cent. Thus, if ${\gamma }_{KZ}^2<2$ , the marginal revenue from quality upgrading is larger than the MC following a 1 per cent increase in the rate of per unit production subsidy, so that producers are incentivised to raise quality.

4.5 Role of Industrial Policy under Asymmetric Information

In Section 3 we discussed how quality being unobserved by consumers and consequent asymmetric information lead to moral hazards and adverse selection problems, and how, accordingly, bad quality drives good quality out of the market (the so-called lemons problem). Since consumers associate the quality of a good they consume with its country of origin, one policy that works in favour of potentially high-quality producers in a country for which consumers have poor country-of-origin perception is having export-quality standards and quality certifications for the products of individual exporting firms.

Quality labelling, like ISO 9000 and ISO 14000 certifications, signals to consumers that the product has attained a minimum quality standard. A potentially high-quality producer in a developing country can thus get a higher price than she would get based on buyers’ perceptions of the average industry quality. Developing countries and companies therein increasingly use management and ISO 9000 certifications to overcome reputation problems and signal their investment into quality upgrading and quality performance (Reference Hudson and JonesHudson and Jones, 2003; Reference Potoski and PrakashPotoski and Prakash, 2009; Reference FerroFerro, 2011; Reference Montiel, Husted and ChristmannMontiel, Husted and Christmann, 2012). Despite the potential advantage of overcoming the negative informational externality problem arising from poor average industry quality, obtaining ISO certificates from certifying bodies is quite costly. Such costs involve not only application and process fees but also organisational restructuring and product standardisation costs, which are often not affordable for small producers in developing countries. In fact, these costs are usually higher in developing countries than in developed countries and can be a barrier to trade (Reference Maskus, Otsuki and WilsonMaskus, Otsuki and Wilson, 2005; Reference Clougherty and GrajekClougherty and Grajek, 2008, Reference Clougherty and Grajek2014; Reference Auriol and SchilizziAuriol and Schilizzi, 2015). This necessitates government subsidies, which are also socially optimal since the social marginal benefit of obtaining such certificates and thereby reducing information externality is much higher than the private marginal benefits.

There is, however, contrasting evidence on the effect of ISO certifications. A survey on exporting firms in Bangladesh conducted by Reference Raychaudhuri, Acharyya, Marjit, Bhattacharyya, Rahman and TipuRaychaudhuri et al. (2003) reveals that 36 per cent of these firms experienced an increase in their domestic as well as export sales after obtaining ISO 9000 certification. Expectedly, Reference Clougherty and GrajekClougherty and Grajek (2008) find ISO certification to be more beneficial for developing countries than developed countries. However, Reference Hudson and JonesHudson and Jones (2003) observe that use of ISO may not fully eliminate informational asymmetry. One reason is that consumers in developed countries often rely more on domestic standards than on ISO 9000 certifications. This is not only because they are more familiar with the domestic standards but also because of the way ISO 9000 accreditation is organised. On the other hand, the study by Reference Dunu and AyokanmbiDunu and Ayokanmbi (2008) on the impact of ISO 9000 certification on an organisation’s financial performance reveals that such certification does not significantly increase the ratio either of revenue to assets or of operating income to assets. Subsequent studies have observed that the beneficial effect of ISO 9000 certification is contingent upon many institutional factors. For example, Reference Sanetra and MarbanSanetra and Marban (2007) and Peuckert (2014) argue that signalling unobserved quality characteristics through ISO certification depends on the credibility of the certifying institutions, which they termed ‘quality infrastructure’. On the other hand, Reference Sung and ReinertSung and Reinert (2009) identify institutional capacity as the ability of governments and private entities to deliver essential services and develop original measures of four dimensions of standards-related institutional capacity, namely information, conformity, enforcement and international standard-setting. They incorporate these measures into a gravity model to assess their capacity in offsetting the negative effects of standard and technical regulations such as Aflatoxin B1 on food and agricultural product trade. More recently, Reference Blind, Mangelsdorf and PohlischBlind, Mangelsdorf and Pohlisch (2018, p. 50) argue that the impact of ISO certification signalling quality upgrading depends on the ‘trust in the accreditation system and the development status of a country’. The signalling power of certificates is stronger when certification bodies are recognised by accreditation bodies that are signatories to the Multilateral Recognition Arrangement of the International Accreditation Forum (IAF MLA). Reference Blind, Mangelsdorf and PohlischBlind et al. (2018) used a gravity model for the period 1999–2012 to show that certification promotes trade and that signatories to the IAF MLA trade significantly more. This result underlies the importance of support for accreditation institutions in developing countries. The reputation of the accreditation bodies increases if they are recognised through the IAF MLA, which in turn strengthens the signalling effects and accordingly boosts trade.

That certification by independent agencies or third parties is an effective way for producers of high-quality products to distinguish themselves from the industry average and disseminate product-quality signals to consumers has been emphasised by some recent studies as well (Reference Van Loo, Caputo, Nayga, Meullenet and RickeVan Loo et al., 2011; Reference Janssen and HammJanssen and Hamm, 2012; Reference Bruschi, Shershneva, Dolgopolova, Canavari and TeuberBruschi et al., 2015; Reference Elfenbein, Fisman and McManusElfenbein, Fisman and McManus, 2015). Reference Ortega, Wang, Wu and OlynkOrtega et al. (2011) point out that one of the reliable sources of such information are the results of food-safety sampling tests published by the government. The effect of the government’s disclosure of food-quality and food-safety information on the food-quality performance of suppliers is investigated in quite a few studies. For example, evaluating the impact of the hygiene quality grade cards policy in Los Angeles in the USA, Reference Jin and LeslieJin and Leslie (2003) find that the hygiene quality of restaurants improved significantly. Similarly, Reference Ollinger and BovayOllinger and Bovay (2020) observe that both the credible threat and actual actions of public disclosure of chicken quality and safety by the government led to improvement in quality and safety by chicken slaughterers.

The optimal quality certification in terms of the power of certification in motivating producers to improve the quality of their products has, of late, been studied theoretically in a moral hazard setting by Reference ZapechelnyukZapechelnyuk (2020). He shows that when a monopolist chooses quality and price of an indivisible good, simple certification systems like quality assurance and pass-fail rules are optimal for the regulator that maximises consumers’ surplus.

An interesting dimension of quality certification is online marketplaces with quality certifications/badges for sellers. Online marketplaces often use seller reputation scores and badges to certify that sellers meet some minimum quality threshold. Examples are the ‘Top Rated Seller’ badges used by eBay and the labels ‘Superhost’, used by Airbnb, and ‘Top Rated’, used by Upwork. In terms of a simple asymmetric information model, Reference Hui, Saeedi and TadelisHui, Saeedi and Tadelis (2019) show that the average quality of both badged and unbadged sellers increases under more stringent certification. Moreover, certification induces entry by both the highest-quality and the lowest-quality sellers. Using the predictions of their model and studying the differential impact of the policy change across 400 separate subcategories of the eBay marketplace, Reference Hui, Saeedi and TadelisHui et al. (2019) observe that, after the policy change, entry increases more in markets where the fraction of badged sellers fell relatively more, and that the average quality of entrants increases significantly. However, using the reputation scores of online sellers to ascertain quality may make consumers trust certification more than their own information set. And, when certification is not a very transparent procedure and its verification is costly, it may lead to herd behaviour among consumers.Footnote ³²

Another policy that may be effective is improving the country-of-origin perception itself by raising the average industry quality. As Reference Chiang and MassonChiang and Masson (1988) demonstrate, one way to achieve this is by pursuing a policy of industrial consolidation; another is by issuing export licences to only a few firms. These policies essentially internalise the informational externality and encourage exporting firms to upgrade the quality of their products. A very large number of domestic firms with unknown brands lower the average industry quality and adversely impact country-of-origin perception. Reference Chiang and MassonChiang and Masson (1988) cite the example of about 1,000 highly competitive cotton spinning and weaving firms in Taiwan exporting their little-known individual brands in international markets under the label ‘Made in Taiwan’. Low export quality is often related to such a highly fragmented industrial structure, necessitating industrial consolidation. Similar in spirit, Reference Donnenfeld and MayerDonnenfeld and Mayer (1987) demonstrate that a voluntary export restraint can also achieve the socially optimum quality. These arguments, however, rest on the implicit assumption that consumers can correctly assess the average industry quality. It is then obvious that allowing only a single firm to operate in the industry will eliminate all informational externality. But this implicit assumption contradicts the basic premise that quality is unobservable. When industrial concentration is achieved at the extreme with only one firm being given the licence for production and export, say, then the assumption that consumers can correctly perceive the average industry quality means that they can observe the quality of the good that this single firm produces.

4.6 Concluding Remarks

From the discussions in this section emerge a few important lessons for policymakers in developing countries targeting export promotion through upgrading the quality of export products. First, any policy, whether trade and exchange rate liberalisation or allowing factor flows, may cause asymmetric variations in the quality of export goods depending on those goods’ relative skill intensities for upgrading quality. Second, different policies have contrasting effects on the quality of a particular type of export good because of their asymmetric effects on the capital cost – and, correspondingly, on the skilled-labour cost – of producing higher quality. Accordingly, while pursuing trade policies that promote exports through upgrading their quality, the policymakers of developing countries must take into account the relative skill intensity of the targeted export products.

Third, policies must be targeted instead of being the same across the board. In cases where across-the-board effects of policy changes cannot be avoided – such as when across-the-board tariff reductions are undertaken by a developing country under its commitment as a WTO member, or when the domestic currency is devalued – then those policies must be supplemented by targeted subsidies, such as production subsidies linked to quality level being provided to producers of adversely hit export goods. Input subsidies may also work, although this depends on their nature.

Last but not least, in the case of goods with unobservable qualities, quality labelling, like ISO 9000 and ISO 14000 certifications, to overcome information asymmetry and reputation problems may be an effective solution. But this calls for providing subsidies to exporting firms since such certifications involve high application and processing fees, as well as costly organisational restructuring and product standardisation.

5.1 Trade Liberalisation, Quality Variations and Wage Inequality

Motivated by the observed rising trends in wage inequality almost universally across the globe since the 1980s, a large literature has developed over the last three decades that identifies different channels through which liberal trade policies may accentuate wage inequality in both developed and developing countries (see, e.g., Reference DavisDavis, 1996; Reference Feenstra, Hanson, Feenstra, Grossman and IrwinFeenstra and Hanson, 1996; Reference Marjit and AcharyyaMarjit and Acharyya, 2003, Reference Marjit, Acharyya, Rajan and Reinert2009; Reference XuXu, 2003; Reference Marjit and KarMarjit and Kar, 2005; Reference Zhu and TreflerZhu and Trefler, 2005; Reference Chakraborty, Marjit and YuChakraborty and Sarkar, 2008, Reference Chakraborty, Sarkar, Basu, Chakravarty, Chakrabarti and Gangopadhyay2010; Reference Chaudhuri, Yabuuchi, Marjit and YuChaudhuri and Yabuuchi, 2008; Reference AnwarAnwar, 2009). However, none of these have accounted for the quality variations of skill-based exports induced by export-promotion policies as a plausible source and cause of the observed rise in wage inequality.

One of the very few analyses demonstrating that quality upgrading of skill-based exports causes wage inequality to worsen is that of Reference DasDas (2003). In a two-country model he establishes that free trade improves the quality of the traded good through a pro-competitive effect. This in turn raises the skilled wage because higher qualities are more skilled-labour intensive. The relative wage, however, rises in both the trading countries if they are similar with respect to technology and labour endowment. Otherwise, in a North–South context with the North having more skilled labour absolutely as well as relative to the endowment of unskilled labour in the South, the relative wage increases in the North, whereas it may increase or decrease in the South.

Reference Ma and DeiMa and Dei (2009) have also analysed the impact of quality upgrading on wage inequality. But in their theoretical model, where quality improvement requires more intensive uses of both an imported input and skilled labour, a developing country has an absolute cost disadvantage in all feasible quality levels due to its technological inferiority. This assumption rules out the possibility of intra-industry trade in the quality-differentiated good, unlike Reference Flam and HelpmanFlam and Helpman (1987) or Reference Falvey, Kierzkowski and KierzkowskiFalvey and Kierzkowski (1987). In Reference Ma and DeiMa and Dei (2009), a developing country can produce some low-quality goods only under an ad valorem tariff on imports of quality-differentiated goods. These different qualities that can be produced domestically under a tariff wall, however, cannot be exported and are essentially non-traded. In such a context, Reference Ma and DeiMa and Dei (2009) show that a reduction of the input tariff improves the quality of domestically produced non-traded varieties and worsens wage inequality.

The quality upgrading of differentiated export goods induced by tariff reduction on the imported input used in the production of other export goods can also worsen wage inequality. To exemplify, consider the extended A–J framework. As shown in Section 4, a tariff reduction on an imported input used by a composite traded good raises the effective domestic price of the composite traded good: ${\tilde{P}}_T^W=P_T^W-a_{IT}(1+\tau )P_I^W$ . At initial quality of the skill-based export good Z, this raises the rate of return to capital and lowers the unskilled wage by the standard price-magnification effect since the non-traded good is unskilled labour intensive relative to the composite traded good. Higher capital cost forces producers of good Z to lay off skilled workers and contract production. Competition among skilled workers lowers the skilled wage to the level that allows the producers to break even. Thus, at the initial level of quality of Z, input-tariff reduction lowers both the skilled and the unskilled wages. There will, however, be secondary effects on wages through changes in the quality of Z induced by this tariff reduction. Recall that input-tariff reduction improves quality if higher qualities are relatively skill intensive ( ${\gamma }_{SZ}>{\gamma }_{KZ}$ ), and worsens it otherwise. However, regardless of whether quality is upgraded or downgraded, the demand for capital in Z sector will fall. The reason is simple. Suppose that ${\gamma }_{SZ}>{\gamma }_{KZ}$ so that input-tariff reduction improves the quality of Z. This raises the demand for capital by ${\gamma }_{KZ}$ at the margin. On the other hand, larger demand for skilled labour for the higher quality lowers the output of Z, due to the specificity of skilled labour, by the magnitude ${\gamma }_{SZ}$ at the margin. On account of this scale effect, the demand for capital falls. Since ${\gamma }_{SZ}>{\gamma }_{KZ}$ , overall, the demand for capital in this sector falls. In the opposite case, where ${\gamma }_{SZ}<{\gamma }_{KZ}$ , export quality is downgraded consequent upon input-tariff reduction, which lowers the demand for capital by ${\gamma }_{KZ}$ and for skilled labour by ${\gamma }_{SZ}$ at the margin. The latter causes a scale expansion requiring a larger amount of capital in Z sector. But by ${\gamma }_{SZ}<{\gamma }_{KZ}$ , this scale effect is now smaller and thus, again, the overall demand for capital in Z sector falls.Footnote ³⁶

Thus, regardless of the nature of the quality variation induced by an input-tariff reduction, capital availability for the (T, N) nugget rises. By the standard output-magnification effect, this raises output of the composite traded good and lowers that of the non-traded good since the former is relatively capital intensive. Consequently, the price of the non-traded good rises to clear the local market, which, in turn, raises the unskilled wage and lowers the rate of return to capital. The skilled wage rises, too, as a consequence. Though these changes do not reverse the initial declines in the wages and the rate of return to capital, the central point is that quality variations – or, more specifically, quality upgrading – raise the unskilled wage, which minimises the deteriorating absolute position of unskilled workers.Footnote ³⁷ However, since the skilled wage rises too, it may be worthwhile examining whether wage inequality worsens on account of quality variations or not. Algebraically, overall changes in the two wages decomposed into primary and secondary effects can be worked out as:

$\widehat{w}=\frac{-{\theta }_{IT}{\theta }_{KN}\gamma \widehat{\tau }+{\theta }_{KT}\left[\frac{{\lambda }_{KZ}({\gamma }_{SZ}-{\gamma }_{KZ})}{{\epsilon }_N\left|\lambda \right|}\right]\widehat{Q}}{\left|\theta \right|}$(46)

${\widehat{w}}_S=\frac{-{\theta }_{KZ}{\theta }_{IT}{\theta }_{LN}\gamma \widehat{\tau }+{\theta }_{KZ}{\theta }_{LT}\left[\frac{{\lambda }_{KZ}({\gamma }_{SZ}-{\gamma }_{KZ})}{{\epsilon }_N\left|\lambda \right|}\right]\widehat{Q}}{{\theta }_{SZ}\left|\theta \right|}$(47)

where ${\epsilon }_N$ is the absolute value of the price elasticity of relative demand for the non-traded good; ${\theta }_{ij}$ is the share of the i-th factor in the unit cost of producing the j-th good; ${\lambda }_{KZ}$ is the share of capital used in the Z sector; $\left|\lambda \right|$ and $\left|\theta \right|$ are, respectively, determinants of employment-share and cost-share matrices (for the (T,N) nugget); and $\gamma \equiv \frac{\tau }{1+\tau }$ .

The second terms in the numerator of both these expressions capture the effect of quality variations induced by the input-tariff reduction. Subtracting one from the other and using the expression for quality change gives us the change in wage inequality consequent upon quality variations:

${\widehat{w}}_S-\widehat{w}=\frac{{\theta }_{KZ}{\theta }_{IT}{\theta }_{LT}{\theta }_{SZ}{\theta }_{LN}{\lambda }_{KZ}\gamma \alpha }{\delta \mathrm{\Delta }{\theta }_{SZ}{\epsilon }_N\left|\lambda \right|\left|\theta \right|}\left[\frac{{\theta }_{KZ}}{{\theta }_{SZ}}-\frac{{\theta }_{KT}}{{\theta }_{LT}}\right]{({\gamma }_{SZ}-{\gamma }_{KZ})}^2\widehat{\tau }$(48)

Since, as explained earlier, $\widehat{Q}>0$ if ${\gamma }_{SZ}>{\gamma }_{KZ}$ , so quality variations worsen wage inequality $({\widehat{w}}_S-\widehat{w}>0)$ if

$\frac{{\theta }_{KZ}}{{\theta }_{SZ}}>\frac{{\theta }_{KT}}{{\theta }_{LT}}$(49)

Note that, given the technological specifications in Equations (28) and (29), $\frac{{\theta }_{KZ}}{{\theta }_{SZ}}$ rises with the quality level if higher-quality varieties are relatively more capital intensive, and falls with the quality level otherwise. More precisely, let $\mu (Q)\equiv \frac{{\theta }_{KZ}(Q)}{{\theta }_{SZ}(Q)}.$ It is straightforward to check that ${\mu }^{\prime }(Q)>0\forall Q\in [0,\overline{Q}]$ if ${\gamma }_{KZ}>{\gamma }_{SZ}$ , and ${\mu }^{\prime }(Q)<0\forall Q\in [0,\overline{Q}]$ otherwise. Figure 10 illustrates these alternative cases.Footnote ³⁸ In panel (a), the $\mu (Q)$ curve representing the cost-share ratio on the left-hand side of the inequality in Equation (49) is drawn upward sloping for ${\gamma }_{KZ}>{\gamma }_{SZ}$ , whereas the horizontal line represents the right-hand-side cost–share ratio of the inequality in Equation (49). If the $\mu (Q)$ curve lies above the horizontal line for the entire feasible range of qualities – as shown by the broken curve in panel (a) – then the condition in Equation (49) is always satisfied and therefore wage inequality worsens following an input-tariff reduction regardless of the initial level of quality. But if the $\mu (Q)$ curve intersects the horizontal line from below – as shown by the solid upward-sloping curve – then the condition in Equation (49) is satisfied and thus wage inequality worsens only when the initial quality of good Z is larger than $\tilde{Q}$ . In panel (b) we portray this sub-case of conditional worsening of wage inequality when ${\gamma }_{KZ}<{\gamma }_{SZ}$ and thus the $\mu (Q)$ curve is downward sloping. In this case, wage inequality worsens if the initial quality of good Z is smaller than $\tilde{Q}$ . To sum up, whether wage inequality worsens or not through quality variations induced by input-tariff reduction depends on the initial level of quality and on the nature of the export product captured through its relative skill intensity.

Figure 10 Initial choice of quality and wage inequality

On the other hand, Reference Ganguly and AcharyyaGanguly and Acharyya (2021) demonstrate how incentivising quality upgrading of a skill-based export good through tariff reduction on an imported final consumption good causes wage inequality to worsen. Recall the analytical structure of a small, open economy producing four goods – two quality-differentiated export goods (Z₁ and Z₂), a homogeneous export good (X) and a homogeneous import-competing good (Y) – discussed in Section 4.1.1. The homogenous export good is produced in the informal sector, such as agricultural products, small-scale textiles and/or leather manufacturing and the like. The sector producing the import-competing good, like manufacturing produced in medium to large-scale business set-ups using unskilled workers employed through formal contracts, is the formal sector. In such a set-up, a ceteris paribus reduction in the tariff rate on the final import good Y will increase foreign competition and force the formal-sector import-competing production to contract. This will release both unskilled workers and K₁-type capital. In addition to this, there will be further displacement of unskilled workers through a change in production technique. As scale contraction releases K₁-type capital, r₁ falls at the initial levels of quality and output of Z₁, which induces producers of Y to substitute unskilled labour with capital per unit of output. As the unskilled workers displaced from the contracting formal sector, by the effects of both scale and technique, move to the informal X sector, competition among them and between them and the incumbent workers there lowers the informal wage. That is, tariff reduction causes the unskilled informal wage to fall through informalisation.Footnote ³⁹

On the other hand, as spelt out in Section 4, the skilled wage will rise unambiguously. Recall, at the initial quality of good Z₁, the lower capital cost induces producers of Z₁ to expand output levels. This raises the demand for skilled labour in Z₁, and correspondingly the skilled wage. Therefore, at the initial quality levels of the two skill-based goods, Z₁ and Z₂, wage inequality rises in three dimensions: between skilled workers and formal unskilled workers (captured through the rise in w_S), between skilled workers and informal unskilled workers (since ${\widehat{w}}_S^{}>0>\widehat{w}$ ) and between both informal and formal unskilled workers themselves (captured through the fall in w). Note that these three dimensions of wage inequality arise because of the segmented markets for unskilled workers, that is, the coexistence of formal and informal labour markets. Destruction of formal jobs in developing countries due to import competition and the informalisation of such economies are discussed in Reference MarjitMarjit (2003), Reference Marjit, Kar and BeladiMarjit, Kar and Beladi (2007), Reference Brady, Kaya and GereffiBrady, Kaya and Gereffi (2011) and Reference BogliacciniBogliaccini (2013). For example, as observed by Reference BogliacciniBogliaccini (2013), in the more industrially advanced Latin American countries, formal industrial employment declined significantly as a consequence of trade liberalisation and there being no alternative sources of formal employment to which the laid-off industrial workers could be relocated. This reduced the income of those unskilled workers, who were pushed into survival strategies in the informal sectors.

Wage inequality may worsen further as tariff reduction raises Q₁ and lowers Q₂. As shown by Reference Ganguly and AcharyyaGanguly and Acharyya (2021), these quality variations increase the capital requirement in the quality-differentiated sectors, thereby causing the formal sector to contract further and the unskilled informal wage to fall even more. To explain, as tariff reduction upgrades quality Q₁, it requires more K₁-type capital per unit of output of Z₁ by the magnitude ${\gamma }_{KZ}^1{\widehat{Q}}_1$ at the margin. On the other hand, the higher demand for skilled labour as Q₁ is raised lowers the production of good Z₁ by ${\gamma }_{SZ}^1{\widehat{Q}}_1$ at the margin. There will be further change in the capital requirement in Z₁ sector induced by the downgrading of Q₂, too. A lower-quality Q₂ requires less skilled labour as well as less K₂-type capital (which is specific to the Z₂ sector) per unit of output. The latter enables production of good Z₂ to expand, which will require additional skilled labour. But, given the assumption that ${\gamma }_{KZ}^2<{\gamma }_{SZ}^2$ , overall demand for skilled labour in Z₂ sector falls. The displaced skilled workers move to Z₁ sector, thereby expanding the production of Z₁. Due to this second-round scale expansion, the demand for K₁-type capital rises. Algebraically, denoting the capital requirement in Z₁ sector by $K_{1Z}^{}=a_{KZ}^1(Q)Z_1$ , we can derive the proportional change in it in terms of quality variations:

${\widehat{K}}_{1Z}={\lambda }_{KZ}^1\left[({\gamma }_{KZ}^1-{\gamma }_{SZ}^1){\widehat{Q}}_1-\frac{{\lambda }_{SZ}^2}{{\lambda }_{SZ}^1}({\gamma }_{SZ}^2-{\gamma }_{KZ}^2){\widehat{Q}}_2\right]>0$(50)

where ${\lambda }_{KZ}^1$ is the share of good Z₁ in the total use of K₁-type capital and ${\lambda }_{SZ}^m$ is the share of good m (m =1,2) in the use of skilled labour.

This increased demand for K₁-type capital will be met by drawing it from the import-competing Y sector, thereby inducing further contraction of the formal import-competing sector. This will release more unskilled workers from there, who will migrate to the informal agricultural sector X, causing a further fall in the informal unskilled wage. Therefore, quality variations induced by tariff reduction accentuate wage inequality further. This result has important ramifications for the existing literature on trade and wage inequality. Wage inequality worsens to a larger extent when quality variations of skill-based exports are taken into account than when these are not accounted for. Thus, the magnitude of changes in wage inequality due to export-promotion strategies like tariff reduction will be underestimated if quality variations induced by such policies are not accounted for. Figure 11, taken from Reference Ganguly and AcharyyaGanguly and Acharyya (2021), illustrates this. Each flatter curve indicates the inverse relationship between the skilled–unskilled wage ratio and the tariff rate for any given pair of (Q₁, Q₂) and corresponding level of $K_{Z1}^1$ , whereas a higher curve reflects larger wage inequality corresponding to a larger $K_{Z1}^1$ , for the reasons spelt out earlier. When the tariff is lowered from t₀ to t₁, at initial quality levels wage inequality accentuates from ${\omega }_0$ to ${\omega }_1{}^{\prime }$ (corresponding to movement from point a to point b). Then, it changes further to ${\omega }_1$ (corresponding to movement from point b to point c on a higher $\varpi$ curve) due to quality variations brought about by the lower tariff. So, taking into account quality variations induced by tariff reduction, the total change in wage inequality should be measured by the movement along the steeper curve.

Figure 11 Effect of tariff reduction on wage inequality

Note that if the informal agricultural sector had used the same capital K₁ as used in the (Y, Z₁) subsystem, tariff reduction would have had the same asymmetric quality effect. But now the informal wage would rise unambiguously despite a tariff reduction causing informalisation, and so the initial or pre-reform quality of export good Z₁ would matter for whether wage inequality worsened or not.

5.2 Devaluation, Quality Variations and Wage Inequality

Currency devaluation has an altogether different implication for wage inequality, which again suggests that not all types of indirect export-promotion policy have the same or similar effects. To see this, consider the benchmark set-up discussed earlier, taken from Reference Ganguly and AcharyyaGanguly and Acharyya (2021), with specific factors (land and capital) used in production in the formal and informal sectors. Devaluation of the domestic currency vis-à-vis the foreign currency generates the following general equilibrium effects.

First, this makes imports dearer in domestic currency terms, causing demand for locally produced import-competing goods to rise. The formal-sector production thus expands. This raises demand for both unskilled labour and K₁-type capital, as well as raising the unskilled wage and the rate of return to the K₁-type capital. Note that the informal unskilled wage rises on two counts. One is due to the increasing marginal productivity as unskilled workers are drawn from the informal sector to meet the increased demand in the expanding formal sector. The other is due to there being factor substitution in the formal sector of capital by unskilled labour in the face of higher r₁ and fixed formal wage. So, at initial quality levels, devaluation unambiguously raises the informal wage through reduced informalisation. In the case of the skilled wage (see Section 4.2), it may fall or rise depending on whether devaluation raises the marginal revenue from quality less or more than the capital (at the margin) cost of raising quality. Algebraically, the change in skilled wage is given as:

${\widehat{w}}_S=\frac{{\theta }_{KY}-{\theta }_{KZ}^1}{{\theta }_{SZ}^1{\theta }_{KY}}\widehat{e}$(51)

Thus, ${\widehat{w}}_S>0$ if ${\theta }_{KY}>{\theta }_{KZ}^1$ . At the initial quality levels, however, wage inequality may decline even if the skilled wage increases. Quality changes induced by currency devaluation will bring in further changes in the informal unskilled wage and, correspondingly, wage inequality. Recall from Section 4 that currency devaluation lowers Q₁ and raises Q₂ and thus by Equation (50) $K_{Z1}^1<0$ . That is, devaluation-induced quality variations release some K₁-type capital, which allows the formal sector to expand. The informalisation falls, further raising the unskilled informal wage. So, quality variations now lower wage inequality in sharp contrast to what we observed under the tariff reduction discussed earlier. Therefore, even if wage inequality is worsened by devaluation at initial quality levels, it may decline at the new equilibrium.

5.3 Uniform Minimum Wage Laws, Quality Variations and Unemployment

Suppose a uniform minimum wage law is implemented in all sectors. That is, there is no informal or organised sector where competition may drive down unskilled wages even below the minimum wage applied in the formal sectors. If such an across-the-board uniform minimum wage is set above the labour market clearing wage, there will emerge an overall unemployment in the economy.Footnote ⁴⁰ In such a context, the issue at hand is whether quality variations will worsen the unemployment situation or not.

In the extended A–J framework, with the wage to unskilled labour pegged at the level $\overline{w}$ by the government in both the composite traded good and the non-traded good, it can be shown that policy-induced quality variations will unambiguously raise aggregate employment. But the overall employment effect depends on the nature of the policy. At the initial quality level, whereas an input-tariff reduction lowers aggregate employment, a devaluation raises it. To check, first of all, recall from Section 4 that under a rigid unskilled wage, a reduction in the input tariff and a devaluation both raise the rate of return to capital, though by different magnitudes. The corresponding increase in the capital cost raises the price of the non-traded good. Demand for and output of the non-traded good thus falls. On this account, some unskilled workers are laid off. Some capital is also released that enables output of the composite traded good to expand, thereby absorbing some of the unemployed unskilled workers. But this good being relatively capital intensive, fewer workers are absorbed than are laid off. Therefore, aggregate employment falls at the initial quality level.Footnote ⁴¹ On the other hand, for the same reason as discussed earlier, input-tariff-reduction-induced quality variation, regardless of its direction, lowers the demand for capital in the Z sector. Larger availability of capital as a consequence raises the output levels of both the composite traded good and the non-traded good (though at different proportions) and thereby raises the aggregate employment of unskilled workers. Overall, the effect of an input-tariff reduction on the aggregate employment is ambiguous. Since the initial adverse effect is driven by a rise in price of the non-traded good and the consequent fall in its demand and output, its magnitude is larger the larger is the price elasticity of (relative) demand for the non-traded good. Accordingly, taking into account any subsequent increase in the employment level due to quality variation, the aggregate employment increases at the new equilibrium if the price elasticity is sufficiently low. The algebraic details are avoided here, but the precise condition or critical value of the price elasticity of (relative) demand for the non-traded good can be easily worked out.

Currency devaluation, on the other hand, also raises the domestic currency price of the composite traded good net of the domestic currency price of the imported input, the effective price, and, by competitive conditions, such a rise must be equal to the increased capital cost. Thus, now the relative price of the non-traded good does not rise unambiguously. More precisely, recalling from Section 4 the magnitude of the proportional rise in the rate of return to capital, $\widehat{r}=\frac{1-{\theta }_{IT}}{{\theta }_{KT}}\widehat{e}$ , the proportional change in the relative price of the non-traded good is given by:

${\widehat{P}}_N-{\widehat{P}}_T=\left[\frac{\left(1-{\theta }_{IT}\right){\theta }_{KN}}{{\theta }_{KT}}-1\right]\widehat{e}=\left[\frac{\left(1-{\theta }_{IT}\right){\theta }_{KN}-{\theta }_{KT}}{{\theta }_{KT}}\right]\widehat{e}$(52)

So, the relative price of the non-traded good falls under the assumption that it is relatively labour intensive (that is, ${\theta }_{KN}<{\theta }_{KT}$ ). Its demand and output thus rise, drawing labour from the unemployed pool and capital from the composite traded sector, necessitating a fall in output of the composite traded good. A fall in employment due to such contraction, however, is less than the additional workers employed in the non-traded sector. Thus, at the initial quality level, currency devaluation raises aggregate employment, unlike input-tariff reduction. Subsequently, quality variations due to currency devaluation will raise aggregate employment for the same reason as spelt out earlier. Thus, the initial employment increase is reinforced by subsequent quality variations.Footnote ⁴²

5.4 Concluding Remarks

While improving export quality and consequent export earnings through indirect export-promotion strategies is a matter of concern for policymakers in developing countries, the adverse consequences of such quality improvement for the domestic unskilled labour markets cannot be overlooked. However, whether such a policy conflict will arise or not is conditional on several factors. For example, when quality variation is induced by an input-tariff reduction, whether the wage inequality worsens or not depends on the initial level of quality and on the nature of the export product captured through its relative skill intensity. Incentivising quality upgrading of a skill-based export good through reduction of tariff on an imported final consumption good, on the other hand, causes wage inequality to worsen unambiguously. By displacing unskilled workers from the contracting formal sector and causing the unskilled informal wage to fall through informalisation, it worsens wage inequality in all three dimensions. Moreover, wage inequality will worsen to a greater extent when quality variations induced by tariff reductions are taken into account. Currency devaluation, however, will have an altogether different implication for wage inequality, which again suggests that not all types of indirect export-promotion policy have the same or similar effects. On the other hand, in economies with prevalence of unemployment of unskilled workers because of implementation of the (uniform) minimum wage laws in all sectors, policy-induced quality variations unambiguously raise the aggregate employment. Still, policy choice may matter because different policies impact aggregate employment in a contrasting manner at the initial quality level. For example, at the initial quality level, whereas an input-tariff reduction lowers aggregate employment, a currency devaluation raises it.

6.1 Domestic Consumption of Quality-Differentiated Export Goods

To keep things simple, consider the case of only one quality-differentiated good, say, Z₁ – and correspondingly only one type of capital in our economy (K₁) – along with the two homogeneous traded goods X and Y. Suppose only skilled workers consume Z₁ along with X and Y. Others consume only X and Y. Let $b_Z(Q)$ be the fraction of skilled wage income, $w_S\overline{S}$ , that is spent on Z₁, where ${b^{\prime }}_Z(Q)>0$ for all $Q\in [0,1]$ but $b_Z(1)={\overline{b}}_Z<1.$

From the marginal condition for choice of quality of Z₁, reproduced below from Section 4, it is evident that Q₁ will be affected by the spending on consumption of Z₁ by the skilled workers only if the skilled wage and the rate of return to capital are affected:

$P_{Z1}^W{}^{\prime }(Q_1^0)=a_{KZ}^m{}^{\prime }(Q_1^0)r_1+a_{SZ}^m{}^{\prime }(Q_1^0)w_S$(53)

But, as is evident from the relevant zero-profit condition, given the rigid unskilled wage in the import-competing sector, the rate of return to capital is tied down by the tariff-inclusive price of imports of Y. This, in turn, ties down the skilled wage for any given Q₁. Moreover, by the envelope theorem, skilled wage remains invariant with respect to changes in the quality level of Z₁. On the other hand, by the small country assumption, a fraction of skilled-wage income now spent on the quality-differentiated export good will only lower the volume of exports of Z₁ without affecting its world price. Further, as long as the rest of the goods are all traded, there will be no changes in the production levels, either. The output levels of the homogeneous traded goods X and Y will not be constrained by local demand and thus will always be at full employment. Hence, the informal unskilled wage will be invariant with the domestic consumption spending on the quality-differentiated export good and correspondingly smaller spending on X and Y. The only change would be the volume of trade. With less of a fraction of income now being spent on good Y (and on X), demand for Y  and thus for imports falls, whereas the volume of agricultural exports rises.Footnote ⁴³ But these changes in trade volumes will be inconsequential for the world prices of X and Y  as long as the country under consideration is small in the world market. Thus, with world prices of all goods – Z₁, X and Y – remaining unchanged, the skilled wage, the rate of return to capital and the unskilled wage all remain unchanged. This no-impact result implies two things. First, the domestic producers of good Z₁ will choose the same level of quality now as when it is not domestically consumed. Second, there will be no change in informalisation and wage inequality.

By similar reasoning, trade and exchange rate policies will have the same effects as we discussed earlier without any domestic consumption of Z₁. Further, whether only skilled workers or all income earners consume quality-differentiated export goods along with the homogeneous traded good does not matter. The results remain unchanged as long as all goods are traded and the economy under consideration is small in the world market.

Domestic consumption matters, however, when the economy produces some non-traded goods. In such a case, production levels and, consequently, factor prices are no longer independent of the local demand conditions, even for a small, open economy. Consider, for example, the A–J framework in which the small, open economy produces a homogeneous non-traded good (N) along with a homogeneous composite traded good (T) and the quality-differentiated export good (Z). Now, domestic consumption of Z means that smaller fractions of income are spent on the composite traded good and, most importantly, on the non-traded good. For any price level, the demand for the non-traded good falls. The consequent excess supply lowers its price, which in turn lowers the unskilled wage and raises the rate of return to capital by the price-magnification effect as long as the non-traded good is unskilled labour intensive relative to the composite traded good. The skilled wage, on the other hand, falls for reasons spelt out earlier. Thus, by the marginal condition for quality, the quality of the export good Z will rise if higher qualities are more skill and labour intensive ( ${\gamma }_{SZ}>{\gamma }_{KZ}$ ). Wage inequality will also change as a consequence of domestic consumption, but its direction will be ambiguous since both the unskilled and the skilled wages decline.

On the other hand, if there are uniform rigid wages causing aggregate unemployment of unskilled workers, then, though domestic consumption of Z will leave its quality unchanged, the aggregate employment will fall unambiguously. Note that, similar to what has been discussed already, with the unskilled wage being rigid, the rate of return to capital is tied down by the world price of the composite traded good and this, in turn, ties down the skilled wage for any given Q₁. With the quality of Z thus remaining unchanged, its output remains the same as well. Hence, there will be no change in the demand for capital in that sector or, correspondingly, in the capital available for the (T, N) nugget, leaving the output levels of the traded and the non-traded goods unchanged. But, since the domestic consumers now spend less on the non-traded (as well as the composite traded) good, so its demand and consequently its output will fall. Released capital on account of this will enable the composite traded good sector to expand. That is, there will be a change in composition of output in the (T, N) nugget away from the non-traded good, causing a fall in the aggregate employment of unskilled labour if it is relatively unskilled labour intensive. The essence of this employment effect of domestic consumption of the quality-differentiated export good Z is similar to a fall in the real income of consumers: it is as if we tax away the part of domestic consumers’ income that they spend on T and N.

6.2 Monopoly Production

Suppose Z₁ is produced by a single firm, whereas Z₂ is produced by perfectly competitive firms. As a further simplification, suppose the domestic firm charges a fixed mark-up $\varphi$ over the average factor cost for any given Q₁, which she raises, however, with the quality level: $\varphi$ (Q₁), ${\varphi }^{\prime }$ > 0. Since the domestic economy is small in the world market, the market power of this domestic firm is restricted only within the domestic economy. So for any quality level Q₁ that she chooses, the monopolist cannot charge a price higher than $P_{Z1}^W(Q_1)$ , such that

$P_{Z1}^W(Q_1)=\Phi (Q_1).C_{Z1}(Q_1)=\Phi (Q_1)\left[a_{SZ}^1(Q_1)w_S+a_{KZ}^1(Q_1)r_1\right]$(54)

where $\Phi (Q_1)=1+\varphi (Q_1)>1$ .

For any given mark-up and choice of Q₁, the domestic firm pays a wage to the skilled workers according to Equation (54). Of course, she pays less if she raises her mark-up and vice versa. The quality is chosen by the following marginal condition:

$\frac{d{\pi }_{Z1}}{d{Q}_1}=P_{Z1}^W{}^{\prime }(Q_1)-C_{Z1}(Q_1).{\Phi }^{\prime }(Q_1)-\Phi (Q_1).C_{Z1}{}^{\prime }(Q_1)=0$(55)

First, it can be easily verified that the monopolist will provide a lower quality for any given skilled wage and rate of return to capital, which follows from evaluating Equation (55) at Q₁ produced by the competitive producers that satisfy the marginal condition in Equation (53):

${\frac{d{\pi }_{Z1}}{d{Q}_1}|}_{Q_1^0}=[1-\Phi (Q_1^0)]P_{Z1}^W{}^{\prime }(Q_1^0)-C_{Z1}(Q_1^0).{\Phi }^{\prime }(Q_1^0)<0$(56)

This under-provision of quality result, illustrated in Figure 12, is similar to that established in the partial equilibrium quality choice models discussed in Section 3. The upward-sloping convex curve labelled $P_{Z1}^W(Q_1)$ represents the MWP of consumers in the world market, which increases with the quality level of the export good Z₁ at the increasing rate. On the other hand, the upward-sloping convex curve labelled $C_{Z1}(Q_1,w_S,r_1)$ represents the unit cost of producing good Z₁, which, by the cost-specifications in Section 3, increases at an increasing rate with the quality level Q₁ for any given w_S and r₁. Competitive firms choose quality level Q^C₁ corresponding to the tangency between these two curves, so that both the zero-profit condition and the marginal condition are satisfied. The monopolist producer’s profit-maximising quality choice is shown by the tangency between a lower unit cost curve – since it pays lower wages to the skilled workers – and the curve labelled $\frac{P_{Z1}^W(Q_1)}{\Phi (Q_1)}$ , satisfying the conditions in Equations (54) and (55). Note that the vertical distance between the $P_{Z1}^W(Q_1)$ and $\frac{P_{Z1}^W(Q_1)}{\Phi (Q_1)}$ curves represents the mark-up or profit per unit of output for the monopolist.