Globalization and Perceptions of Upward Mobility in Developing Countries: An Experimental Study*

Transcription

1 Globalization and Perceptions of Upward Mobility in Developing Countries: An Experimental Study* Sera Linardi and Nita Rudra *Preliminary Draft Do citizens demand more redistribution as their economies become increasingly exposed to international market competition? The link between globalization and the demand for redistribution has never been established in an empirical test. A central contention in the international and comparative political economy literatures is that governments facilitate redistribution policies alongside international market expansion in order to reduce worker insecurities, and buy support for openness policies. However, the causal mechanisms linking globalization and why citizens might, or might not, demand redistribution have not been fully explored. It is critical to assess whether globalization actually generates greater public support for redistribution to confirm that such a link exists. In doing so, we connect with a wealth of literature in economics which explores micro-level factors affecting demand for redistribution, heretofore neglected in the globalization-redistribution debate. This analysis investigates the extent to which citizens of developing countries support redistribution in response to a particular dimension of globalization: foreign direct investment (FDI). We focus on FDI rather than trade because it allows for a better test of subjective citizen reactions to globalization in less developed countries (LDCs). FDI often takes the form of international brands and companies which function as highly visible yardsticks of globalization to ordinary citizens of LDCs. Put simply, the presence of multinationals represents a clear 1

2 departure from old development models. Additionally, as Scheve and Slaughter (2004) cogently argue, FDI is a distinguishing characteristic of the current phase of globalization which plays a substantial role in exacerbating worker insecurities. They speculate- but do not test- that this insecurity drives higher demands for redistribution. We explore the link between FDI and individual support for redistribution in developing economies for a variety of reasons. First, the answer provides a missing piece of the puzzle on why LDC governments have been lowering welfare spending concomitant with market expansion. The existing literature finds that unlike advanced industrialized economies, LDC governments are minimizing such expenditures in order to lower labor costs and improve competitiveness in a global economy (Mosley 2003; Kaufman and Segura 2001; Garrett 2001; Nooruddin and Simmons 2009; Rudra 2002, 2008; Segura-Ubiergo 2007; Wibbels 2006; Wibbels and Arce 2003). Yet they are relatively silent on government political rationale and seeming immunity to the political consequences of retrenchment. We are motivated to explore why governments in LDCs feel reducing or ignoring redistribution in a globalizing economy is safe political strategy. Second, assessing determinants of micro-level demand for redistribution in LDCs is a grossly neglected area of research. Since current analyses on this subject focus overwhelmingly on developed economies, this study helps fill an enormous lacuna in the literature. Finally, assessing opportunities for and constraints to redistribution is absolutely critical in these nations given slow and inefficient poverty reduction. Combining insights from behavioral economics and political economy, we predict that FDI decreases support for redistribution in developing economies. Scholars in political economy have recently focused on the prospect of upward mobility (POUM) as a major determinant of demand for redistribution. According to POUM, support for redistribution 2

3 decreases as individuals anticipate gains in their future standard of living. We build on this scholarship and identify how and why FDI influences perceptions of increased opportunity and upward mobility in developing economies, particularly for the poor. As a result, citizensparticularly the more politically vocal better off groups- tend to resist redistribution. We employ both cross-country surveys and experimental evidence to support our theoretical intuition. Our findings represent a clear departure from existing research which assumes globalization generates insecurities that automatically translate into greater demands for redistribution. In doing so, we are the first to empirically identify factors that can affect POUM, and directly impact support for pro-poor policies as a consequence. Existing Literature: Globalization and Support for Redistribution in LDCs Previous empirical research has focused primarily on how globalization- - trade in particular-- impacts redistribution in developed economies (Garrett 1998, Dreher, Strum, and Ursprung 2008, Hays, Ehrlich, and Peinhardt 2005, Burgoon 2001, Mahler 2004, Ehrlich 2010, Brady, Beckfield, and Seeleib-Kaiser 2005, Swank 2005, Ha 2008), although interest in the globalization-redistribution nexus in developing nations has been burgeoning (Haggard and Kaufman 2008, Nooruddin and Simmons 2006, Brooks 2010, Ahlquist 2006, Rudra 2002, 2008; Segura 2007; Wibbels 2006; Wibbels and Arce). In all these studies, it is explicitly or implicitly reasoned that globalization will raise public demands for redistribution because worker insecurities increase concomitant with international market exposure. Scheve and Slaughter (2004) have recently turned scholarly attention towards the links between FDI and demands for redistribution. Using panel data from Great Britain collected from a survey, Scheve and Slaughter (2004) show that FDI exacerbates worker insecurities because: (1) FDI increases labor demand elasticity since multinationals can 3

4 substitute foreign workers with domestic labor; and (2) multinationals (MNCs) can substitute foreign factors of production in response to higher wages. Scheve and Slaughter (2004) speculate that since FDI is positively correlated with individual perceptions of economic insecurities, demand for redistribution will likely increase. We question whether this link between international market-induced worker insecurities and support for redistribution is a foregone conclusion, particularly since it has never been put to an empirical test. Although the causal relationship is intuitive, a wealth of literature in economics explores micro-level determinants of redistribution and provides us with ample reasons why this may not be the case. This scholarship identifies factors such as gender, wealth, age, ideology and number of children to help explain variation in citizen support for redistribution (e.g., Svallfors 1997, Finseraas 2008, Alesina and Ferrara 2005, Rehm 2005, Linos 2003, Alesina and Giuliano 2009). Findings from these studies alert us to the possibility that insecure workers employed by multinationals (or otherwise) may not demand higher redistribution because of individual-level characteristics. Surprisingly, this large body of research has been overlooked by most political scientists focused on explaining redistribution policies. 1 To evaluate competing hypotheses for the causes of redistribution support in a global economy, we explore global explanations from the international political economy (IPE) literature, as well as the scholarship that analyzes micro-level factors driving individual demands 1 This scholarship mostly focuses on how political institutions aggregate preferences towards redistribution. Variables commonly associated with increased support for redistribution are democracy (Meltzer Richard 1981 ) especially in developing countries with low median incomes(nelson 1999), majoritarian systems (Iverson, Torben and David Soskice 2006 ), and political partisanship (Cusack, Iversen, and Rehm 2006). None of these factors, however, shed much light on if and under what conditions rising worker insecurity associated with FDI actually increases demand for redistribution 4

5 for redistribution. We ultimately combine these two literatures to develop a testable hypothesis on the micro-level conditions under which FDI impacts on support for redistribution in developing countries. Our argument has implications for how FDI affects insecurities of workers and non- MNC workers (i.e, unemployed, workers for domestic industries, business owners) alike. FDI and Support for Redistribution in LDCs Redistribution policies can serve two objectives: it can provide social insurance (such as social security, unemployment compensation), and/or redistribute income from the wealthy to the poor (e.g., food subsidies, and other targeted programs for the poor), although some programs may have components of both. Our analysis emphasizes support for pro-poor redistribution by higher-income groups. If risk-averse workers are concerned about future income loss because of globalization, external shocks, etc., they may be less likely to resist propoor policies.. We stress the policy choices of the better-off since this group has the biggest impact on redistribution decisions in LDCs (International Labor Organization 1999, Ross 2006, Rudra 2010). Our theoretical foundations for how FDI affects individual demands for redistribution rest on a particular finding in this literature regarding the role of individual perceptions of social mobility, or more specifically prospects for upward mobility (POUM). Our prediction is that support for redistribution in developing countries will fall as multinationals serve as powerful symbols of progress, and citizens begin to associate FDI with greater prospects for social mobility. More individuals embrace the view that individual, and not government, effort in the near future will help keep poverty at bay. 5

6 The key feature of POUM is the intertemporal nature of demand for redistribution; support for redistribution depends not only on their current income, but their expected future income as well. POUM posits that individuals will oppose high rates of redistribution in anticipation that they (or their offspring) will climb the income latter in the near future (i.e, 1-5 years generally). Theoretical and empirical evidence for this proposition is strong, ranging from formal models (Benabou and Ok 2001, Piketty 1995), survey data (Ravallion and Lokshin 2000, Kaufman 2009, Alesina and LaFerrara 2005, Alesina and Guiliano 2009), and experimental studies (Checchi and Filippin 2004, Beckman 2006, Neudstadt and Zweifel 2009). We anticipate that the presence of foreign firms proxies as information about mobility, and encourages individuals of different ideological persuasions to believe that poverty in the near future will be the result of hard work and effort, rather than luck. The luck versus effort dichotomy is the root of individual s ideological differences on support for redistribution (e.g., Fong 2001, Krawczyk 2010, Ravallion and Lokshin 2000, Alesina and Angeletos 2005). Numerous studies show that when individuals perceive that poverty is a function of factors beyond individual control, such as bad luck, random chance and/or the absence of opportunity, they are more likely to support redistribution (i.e., principle of inequality aversion). On the other hand, when individuals perceive that poverty is a function of (the lack of) effort then they are less likely to support redistribution (i.e, fairness principle). However, under certain conditions, individuals behavior towards redistribution can run counter to their ideological predispositions. More specifically, individuals can make redistribution choices that are counter to their ideological predisposition when they are faced with greater information about circumstances affecting the recipient s poverty situation. For example, liberal voters, who generally believe poverty is a function of bad luck, might decide 6

7 against redistribution when given information that individuals are responsible for their predicament (see Fong 2001, Alesina and Angeletos 2005, Fong and Oberholzer-Gee 2011). By the same token, those who associate poverty with a lack of effort can also decide to support redistribution-type policies after they are given information that the chance of random loss is high (see Esarey, Salmon and Barrilleaux 2011). In LDCs, citizens of different ideological persuasions may observe the growing presence of foreign firms as information that greater individual effort is likely to pay off in the near future. This view may be embraced irrespective of whether the poor have greater actual income opportunities. In other words, the existence of multinationals can represent a powerful symbol of mobility in LDCs- for both the rich and poor- regardless of whether or not it is true. Research confirms that perceptions of social mobility can affect demands for redistribution, independent of objective measures, or actual mobility rates in the country, (Alesina and LaFerrara 2005, Piketty 1995). 2 Indeed, the existing literature suggests that FDI may not increase objective measures of the poor s mobility. FDI in LDCs is generally correlated with greater economic opportunities for higher-income and skill groups, and on the contrary, can hurt lower income groups by exacerbating inequality ( see, for examples, Ang 2009, Sani et al 2010, Carkovic and Levine 2002, Gorg et al 2007, Aitken, Harrison and Lipsey 1996, Feenstra and Hanson 1997, Gopinath and Chen 2003 ). In a developing country setting, then, donors may rely on impressions of increased job opportunities for the poor from FDI. They may ignore or not seek out information about what these job opportunities actually are in order to maintain moral wiggle room to support the view that the poor no longer needs as much help (see Fong and Oberholzer-Gee 2 More specifically, Alesina and LaFerrara 2005 develop an index using predictions from economic theory that reflects rational measures of expected future income. 7

8 2011). 3 Preliminary surveys conducted for this analysis in India illustrate this point: 62% felt the poor would benefit from the presence of a foreign company. In contrast, only 32% felt that they themselves would directly benefit from the same company; though 60% felt that overall, FDI was good for the economy. This is striking, given that studies have established that FDI is not benefitting the poor in India (Chakraborty and Basu 2002, Ray 2012, Rudra 2012), This tendency to associate FDI with increased prospects of social mobility will be particularly acute in LDCs. First, citizens of developing economies have been facing high levels of social indoctrination about the benefits of FDI in the post-dependency era. As Alesina and Glaeser (2004) argue, social indoctrination about the chances of mobility in a society can lead to over-optimism about social mobility and low demand for redistribution, even if this optimism does not correspond to actual mobility rates. Indeed, in contrast to the dependency era, FDI has been widely marketed as a symbol of progress and the road to poverty reduction by policymakers, Washington pundits, media officials, and international financial institutions. Take for instance recent statements by Zambia s foreign minister, during public discussions with the opposition party. According to a local newspaper, the Zambian minister urged opposition parties not to resist foreign investment because, otherwise, the Zambian economy would not grow to expected levels that could reduce poverty if FDI was not attracted in the country. There is no country that has fought poverty without attracting FDI, so let us not resist and discourage FDI since it is good for us as capital for job creation and technology 3 Experimental studies have shown that in situations that requires altruism, individuals exploit moral wiggle room to justify self-interested actions. Plausible excuses allow an individual to protect his reputation while taking the less generous action (Andreoni and Bernheim, 2009, Linardi and McConnell, 2011) Protecting one s self image while choosing self interest sometimes take the form of information avoidance. An experiment by Dana et al.(2007) show that 40% of subjects in a dictator game prefer to not know the recipient payoffs in order to justify a larger allocation for themselves. 8

9 transfer." (emphasis ours, The Post (Zambia), August 9, 2009). 4 Tolerance for inequality is thus likely to increase as FDI influences subjective perceptions of future mobility for the poor. The second reason for over-optimism about FDI in developing economies is rooted in their historical experience with failed development models that eschewed openness, and coexisted with rampant poverty. Hirschman (1973) noted in his classic piece that poor countries have undergone a series of disasters in terms of economic development. He documents why introduction of wholly new doctrines of development that emphasize income distribution, employment and self-reliance effects optimism and increases society s tolerance for income disparities. Hirschman and Rothschild (1973) explain how this optimism takes root after a nation adopts the new doctrine: As individual s welfare depends on his present state of contentment (or, as a proxy, income), as well as on his expected future contentment (or income). Suppose that the individual has very little information about his future income, but at some point a few of his relatives, neighbors, or acquaintances improve their economic or social position. Now, he has something to go on: expecting that his turn will come in due course, he will draw gratification from the advances of others for a while (Hirschman and Rothschild 1973: 29). It is likely, then, that prior to openness, a greater proportion of citizens supported redistribution in countries with a history of weak economic growth. Faced with low POUM, they are likely more susceptible to the view that the determinants of poverty were unjustified (i.e., due to luck) rather than merit-based. Rather, income mobility may be seen as a result of unreasonable behavior such as bribery, or connections with corrupt government and business 4 The Post. August 9, 2009 Zambian minister urges opposition parties not to resist foreign investment. ). 9

10 officials. Against this backdrop, international market exposure may symbolize progress and a new doctrine of development that promotes widespread views of POUM. Incidentally, Hirschman s (1972) analysis served as the basis of POUM that has been empirically explored in recent times, although, somewhat ironically, not yet in LDCs. It is far from surprising that the relatively sudden influx of FDI would increase confidence amongst citizens in LDCs that potential future economic prospects are good. Indeed, Alesina and Ageletos (2005) comparing the US and Europe, confirm that historical experiences can explain country-specific social attitudes about prospects for social mobility. To summarize, FDI reduces support for redistribution because it represents an information short-cut about growing prospects of upward mobility for the poor. In this environment, even insecure workers in multinationals may believe that their situation -- as well as others economic misfortune--is a function of effort, and consequently resist redistribution policies. Our analysis is unique in that we identify key features of globalization that can affect POUM, whereas previous scholarship views POUM as exogenous. We contend that this FDI effect is peculiar to the developing world. Given the level of social indoctrination about the benefits of globalization, alongside its visible expansion in the form of FDI, we anticipate that negative associations between FDI and colonial imperialism propagated during the Dependency era in some countries more strongly than others- no longer lingers. But this is ultimately an empirical question to which we now turn. Testable Hypotheses and Preliminary Evidence 10

11 We evaluate our hypothesis from four angles, exploring whether (1) FDI reduces the probability of support for redistribution; (2) POUM is one of the underlying causal mechanisms; (3) support for less redistribution in response to FDI is particularly pronounced for the rich (and more politically vocal); and (4) our findings are robust when controlling for time and country fixed effects in a panel setup. For each hypothesis, we estimate a base and full model. Data and Methods Starting from the 1970s, the World Value Survey has interviewed individuals across the world, asking questions about a wide range of issues, including support for redistribution. We use all data available from 1990 to Only two redistribution-related questions are asked repeatedly in threes waves ( , , ), allowing us to include a year/wave dummy to assess the impact of FDI and support for redistribution. For questions available repeatedly across waves, we follow Alesina and Guilano (2009) and use fixed effects model to control for country specific differences. All errors are clustered at the country level. The number of countries vary by wave, and goes from a minimum of 14 to a maximum of 54. Summary statistics on our key independent and dependent variable (support for redistribution) are listed below. Summary Statistics Table 1: Variable Mean Std Dev Min Median Max Obs Countries Individual Level Age Full Employment Male Num. Children Scale of Income

12 Country Level FDI (% GDP) Log GDP/cap Growth GDP/cap Polity Employment to Pop Ratio Trade (% GDP) WVS Survey questions Gov do too little Gov provide for everyone Poverty caused by luck Little chance to escape Eliminate inequality important Competition harmful Note: For the variable Polity, -10 indicates full autocracy and +10 indicates full democracy. Full employment is 1 if the respondent is employed full time, 0 otherwise. Our primary hypothesis is as follows: H1: FDI reduces support for redistribution For H1 we look at two questions from the World Values Survey. The first question is How much is the government doing against poverty? Responses range from for 1 for Too Much to 3 for Too Little. The second question is the question: The government should take more responsibility to ensure that everyone is provided for. Responses range from 1 (strongly disagree) to 10 (strongly agree). We ran an ordered logistic regression with country fixed effects for the first question and an OLS regression with year and country fixed effects for the second. The two (A) columns in Table 2 Panel A reports only the coefficient of interest (FDI and income). Full regression tables and control variables can be seen in the Appendix A. Figure 1 plots the predicted probabilities of % of respondents that think government is doing too much against poverty as a function of FDI investment. 12

13 H1 predicts that the coefficients for FDI in both columns (A) in Panel A will be negative and significant. Regression results for both questions confirm our theoretical intuition. As FDI (% GDP) inflows increase from the 25 th to 75 th percentile, the probability that a respondent reports that the government is doing too much against poverty and that eliminating large income inequalities is unimportant increase by 12% (Figure 1) and 27%, respectively. Individual level controls in the full regression tables in Appendix A are largely consistent with previous studies that have shown High Income and Male to be correlated with decreased support for redistribution. We find that Age and Number of children is correlated with increased support for redistribution. Being employed full time has no effect (results for controls reported in Appendix A). Table 2: WVS Results LDC* Panel A: Support for (1) Cross section (2) Time series pro-poor gov spending Ordered Logit OLS with time and country fixed effects Gov do too little against poverty (1-5) Gov provide for everyone (1-10) (A) (B) (A) (B) FDI (% GDP) *** *** ** *** (0.0142) Income *** *** ( ) FDI x Income ( ) Using Lincom (conditional Sig effect of FDI and income) Sig Pseudo R Observations 40,608 40,608 95,930 95,930 Number of Countries Panel B: Cause of poverty (3) Cross section (4) Cross section and POUM for the poor Logistic regression Logistic regression Poverty caused by luck Very little chance to escape poverty (A) (B) (A) (B) 13

14 FDI (% GDP) ** (0.0263) Income ** * (0.0193) FDI x Income * ( ) Using Nlcom: (conditional Not Sig effect of FDI and income) Sig Pseudo R Observations 22,577 22,577 25,258 25,258 Number of Countries Panel C: Emphasis on Effort (5) Cross section (6) Time Series Ordered Logit Eliminating large inequality important (A) (B) OLS with time and country fixed effects Competition is harmful (A) (B) FDI (% GDP) *** *** *** *** (0.0242) (0.025) (0.0343) (0.0361) Income *** *** *** *** (0.0213) (0.0294) ( ) (0.0136) FDI x Income ** ( ) ( ) Using Lincom (conditional Sig effect of FDI and income) Sig Pseudo R Observations 15,249 15,249 84,762 84,762 Number of Countries *See Appendix A for full tables. Figure 1: Predicted probabilities of % responding that the government is doing too much against poverty (Panel A Question (1) Column (A)) 14

15 Fraction of Respondents th percentile 50th percentile 75th percentile FDI (% GDP) H2: FDI reduces support for redistribution in LDCs because of POUM. We next assess the mechanism through which FDI changes support for redistribution. In particular we examine whether FDI improves perceptions of future income mobility because of subjective perceptions of POUM. First, to confirm earlier studies that FDI does not necessarily introduce more actual opportunities for the poor in LDCs, we regressed objective measures of their future mobility (economic growth, GDP per capita, infant mortality, life expectancy) on FDI. Our results confirm the broader findings of the existing literature (e.g., Ang 2009, Sani et al 2010, Carkovic and Levine 2002): FDI is uncorrelated with any of these measures. See Appendix B. If POUM is a driving mechanism, then FDI should not affect perceptions of current poverty, but future poverty. We focus on two questions regarding attitudes about poverty in Table 2, Panel B. The first question in Panel B (3) is arguably driven by ideology (luck vs. effort) andconcerned with current poverty: Why are people in need? Response is coded as one if the survey taker agrees with People are poor because of an unfair society and 0 otherwise ( People are poor because of laziness ). The second question in Panel B (4) inquires 15

16 about respondents perception of the poor s future economic mobility: In your opinion, do most poor people in this country have a chance of escaping from poverty, or is there very little chance of escaping? The response is coded as 1 if the survey taker agrees with the latter statement and 0 otherwise. If FDI improves POUM of the poor, we would expect FDI to have no effect in Panel B (3) but a negative and significant effect on Panel B (4). We ran logistic regressions with country fixed effects on (3) and (4). As before, Column A of Panel B in Table 2 reports only the coefficient for FDI and Income. As expected, we find that FDI is uncorrelated with causes of current poverty but increases POUM of the poor. As FDI increases, survey responders are less likely to think that the poor has little chance to escape poverty. Panel C investigate further evidence that FDI is correlated with higher tolerance for market forces. The dependent variable for the ordered logit in Model (5) is the response to the question What is the importance of eliminating big income inequalities? which ranges from 1 for Very Important to 5 for Not Important At All. Model (6) in Panel C is an OLS regression where the dependent variable is the degree to which responders agree with the statement Competition is harmful. It brings the worst in people. Response is coded as 10 for complete agreement and 1 for complete disagreement. Model (5) includes only country fixed effects since the question was only administered once in WVS. Model (6) includes both year and country fixed effects. We find that the coefficients on FDI is negative and significant across almost all columns, suggesting that high levels of FDI is correlated with decreasing importance in tackling income inequality and acceptance for a competitive society. 16

17 H3: The effect in of FDI on redistribution attitudes is more pronounced for the rich Finally, in Columns (B) in all three panels, we check to see if a respondent s income level affects their response to FDI by interacting FDI with income level. Consistent with the literature, we expect that the rich will more likely associate FDI with greater opportunity for the poor. If correct, the interaction between FDI and income level should be negative and significant for all regressions in Panel A and B and positive for all regressions in Panel C. We report the coefficient of the interaction term as FDI x Income and denote whether the conditional coefficients are significant immediately below it. 5 These coefficients are plotted again income scale in Appendix C. Our regression results indicate that conditional coefficients for all but Question (4) are significant and negative. This indicate that those with higher income are more influenced by FDI in their support for government spending (Panel A (1) and (2)). Higher income respondent are also more likely to tolerate the inequalities from market exposure (Panel C), possibly because they believe that market exposure benefits all, including the poor. This is reflected by their higher likelihood to report higher POUM of the poor (Panel B (4)). Interestingly, the rich are not more likely than the poor to think that poverty is the function of luck.. Rather, FDI seems to reduce support for poverty-reduction policies for higher income individual, perhaps because they view that effort will pay off in the future in a global economy regardless of the cause of poverty. Finally, while we initially found it surprising that the rich are less tolerant of a competitive economy as FDI increases, this may be because the rich feel more insecure about their assets and wealth in a global market economy. 5 Tests for significance are performed using the lincom command in Stata. 17

18 As a whole, the WVS results in Table 2 show a negative correlation between FDI and preference for redistribution in developing countries. This is reflected in decreased support for government pro-poor program, decreased importance of overcoming large income inequalities, and decreasing view that government has a role in providing for all. At the same time FDI also bring more general changes in attitudes: there is a perception of more opportunities to escape poverty especially among those with higher income, and there is more tolerance for competition. Interestingly, we do not see FDI having any effect on base perceptions that poverty is generally a function of effort or luck. The WVS provides some hints to the mechanism that links FDI in developing countries to changes in attitudes towards redistribution. The significant interaction between income and FDI suggest that these changes in attitudes affect the high income more than the poor, indicating that the rich s perception of economic opportunity from FDI for the poor may be disconnected from the poor s actual experience. This disconnect can be remedied if the rich becomes informed about the poor s actual experiences. However, the experimental literature suggests that the rich would not seek this information even if the costs are low due to strategic information avoidance: the perception of economic progress for the poor provide the rich with the moral wriggle room to avoid redistribution. Causal mechanisms: Experimental Design The WVS only establishes the correlation between our variables of interest but does not allow us to confirm causality. Micro level data on individual attitudes before and after the introduction of FDI is needed to test our hypothesis of the causal mechanisms discussed above. The data should allow us to first check whether the highly visible markers of globalization from 18

19 FDI (such as foreign brand exposure, news of factory opening) increase the perception of economic opportunities for the poor among those who are paying for redistribution. We then would like to investigate whether individuals will pay the small cost of learning about the actual impact of FDI on the poor, and lastly, whether those who are paying for redistribution will reduce their contribution to the poor after exposure to FDI. Since it is costless for an individual to profess their support for redistribution on hypothetical questions, and that not doing so may expose the individual to scorn or being label unaltruistic, we will investigate the aforementioned mechanism in an experiment where individuals earn a compensation for filling in a survey and are then given opportunities to redistribute part of their earnings to another person who earns less than them. We describe the experiment briefly below. The experiment was conducted in three venues: (1) university classrooms in the US; (2) university classrooms in India; and (3) public location in India with random sampling of ordinary citizens. India is good representation of a developing country that has attracted a large amount of FDI, and studies suggest that actual mobility rates for the poor in India are unlikely to increase because of greater foreign investment (Chakraborty and Basu 2002, Ray 2012, Rudra 2012). Conducting this experiment in the US allow us to contrast the framing effect of FDI in developed countries to its effect in developing countries. We begin with university students since they represent the upwardly mobile members of developing countries which are most likely to be politically active. At the start of the experiment, each subjects receive an envelope with experimental instructions stating the following: 6 During the survey imagine that you are a worker who makes Rs ($100) a day. Those who complete their surveys will indeed receive up to Rs.1000 as compensation for their time. The survey provides you 6 US Students were asked to imagine themselves as workers in India. 19

20 an opportunity to donate part of your compensation to a poor person who resides in Bangalore (Person A). You will be presented information for a 4 year period and a budget of Rs 1000/day to help make annual donations decisions for Person A. We will randomly pick one of your 4 donation decisions and implement them on <Day> As a consequence, your take home compensation after the survey will be (Rs.1000 the donation to Person A). As stated in the instructions, the experiment consists of four rounds giving. No identifiable information are collected subjects receive an ID receipt that will be matched to an envelope containing their compensation minus their donation by a staff member that is unaware of the nature or content of the experiment. We took pains to address any concerns of anonymity and concerns regarding whether a real person will receive the donated funds. In the 1 st and 2 nd round subjects face the following decision problem: Year 1: Imagine that you are a worker who makes Rs a day. You can redistribute a part of your income to Person A. Person A resides in Bangalore and makes about Rs. 100 a day. Suppose you choose to give person A Rs. X a day. This means you will earn Rs X a day. How much would you like to give to Person A? I would like to donate Rs. (fill in a number between ) to Person A. In the third round, we introduce new job openings that the low-income person may or may not choose to apply. In the Control group, the middle class workers are told that a local Indian firm will have job openings, while in the FDI group, the workers are told that a foreign company will have job openings. Without knowing what these job opportunities really mean, the middle class worker is then asked again to voluntarily give part their earnings to be distributed to the low-income person(s). The middle class worker can pay a small cost (less than 10% of his earnings) to learn about the cost benefit of the job opportunity. This includes the cost of applying to the laborer job, the probability of being hired, the average income earned at the job, and other 20

21 details. At any point the middle class worker can enter the amount they want redistributed and proceed to the next period. The script for the Control group is given below. The script of the FDI treatment group is identical except that the word Indian is replaced by the word foreign. Subjects within an experimental session are randomly assigned to the Control or the FDI treatment. Year 3: Again you are a worker who makes Rs.1000 a day. As before, you can give a part of your income to Person A. Person A previously made Rs.100 per day, however, this year, a new factory owned by a foreign company is hiring. If hired, Person A is guaranteed new income from in Year 2 to Year 4. Information about the job openings that Person A can apply to at the foreign company is provided in three info sheets marked respectively with Application Cost, Probability of Being Hired, and Income If Hired. Acquiring each piece of information will cost you Rs. 50. In other words, if you open two information sheets, you have spent Rs. 100, and you now only have Rs. 900 of your income left in this period. (1) Your income is: Rs (2) Amount spent on info: the number of info sheets you opened (0-3) * Rs.50 (3) Your remaining income: From your remaining income how much would you like to give to Person B? I would like to donate Rs. (a number between 0 and remaining income) to Person B. In order to isolate how framing job openings as Indian and foreign affects POUM of the poor, the info sheets for both treatment are identical. For example, a subject who purchased Application Cost will learn that cost of job application for Person B is Rs. 50 regardless of which treatment the subject happened to be in. A subject that purchase all info sheets will have complete information to calculate the expected per period earnings of Person B when applying to the job opportunity, which at Rs ($1.25) turns out to be an insignificant increase in Person B s welfare (given previous earnings of Rs.50 or $1). Under complete information, a subject who 21

22 is purely motivated by his beneficiary welfare will not change his redistribution preferences when the job opportunity is mentioned. The final round is identical to the third round. This allows subject s beliefs about the beneficiary s job opportunity to stabilize. This is intended to be an analog of the real world situation where the middle class becomes used to the idea of alternative employment for the poor after FDI is first introduced. Subjects are given a survey at the end of the experiment to elicit the motivation for their giving, the expectations regarding his beneficiary s welfare, and demographic and political information. A sample of the survey questions is given below: o Why did you decide to donate the amount you wrote down in Year 1 to Person B? o Recall that Person B makes Rs.100 /day before Year 3. How much do you think Person B makes after the job openings appear in Year 3? State an amount in Rs. / day. The experimental design generates the following variables: giving in each year / period (P1, P2, P3, and P4) amount of information purchased (P3info), and subject s expectation of beneficiary s earning after the job opportunity is mentioned (expected earning) subjects motivation for giving Our hypothesis is that since FDI represents subjective information about mobility, subjects are unlikely to purchase adequate information about their beneficiary s job opportunities to calculate expected earnings. The mention of a new job opening for the poor will 22

23 indeed induce subject to update their expectation regarding their beneficiary s earnings, but this update will be based on biases and intuition instead of facts. In line with our POUM hypothesis, we expect our Indian subjects in the FDI treatment group to have higher expectation of the poor s new income than the Indian subjects in the Control group. As a result, subjects in the FDI group will reduce their redistribution by a larger amount than subjects in the Control group. Causal mechanisms: Experimental Results from US Classrooms. We ran pilot experiments in two social science classrooms at the University of Pittsburgh: a small graduate microeconomics class (Session 1) and a large undergraduate political science class (Session 2). Subjects are asked to imagine a scenario where they are workers in India who have an opportunity to redistribute to a poorer person who is also in India. The demographic and treatment breakdowns are presented in Table 3. We have not started collecting political affiliation in the first classroom. Table 3: Demographics USA Session 1 (grad) Session 2 (ug) Age Female Treatment : FDI Republicans / Libertarians NA 0.26 Democrats / Green NA 0.58 Independent NA 0.15 N India Session 3 Session 4 The table below present summary statistics of the pilot experiments in the two US classrooms. Randomization appears to work: there is no statistically significant difference between the Control and FDI group before the treatment is applied in Period 2. The graduate 23

24 microeconomics classroom gave around $2 in period 1 and 2 while the undergraduate political science class gave slightly more. Recall that in Period 3, subjects in the Control treatment are told that their beneficiary can apply to new openings at an Indian firm while in the FDI treatment subjects are told that their beneficiary can apply to opening at a foreign firm. Subjects in Session 1 are slightly more curious about the Indian firm (Period 3 Info lookup). There is no difference in willingness to seek out information across treatment in Session 2. On average, subjects purchased one out of the three pieces of information necessary to calculate the expected income of their beneficiary, which was $1.25 (Rs.62.50).This means that subject will have to rely on their intuition and beliefs to form expectation of their beneficiary s income. Subjects indeed overestimate their beneficiary s income and consequently reduce their redistributed amount. The difference between Control and FDI treatment is not statistically significant in Session 1, however, the difference in Session 2 is significant at the 10% level. However, contrary to our hypothesis, US students expected the poor s new earnings to be higher in the case of Indian job openings than in the case of foreign job openings. As a result, even though giving in Period 3 drops significantly from giving in Period 2 in all cases, the drop is less in the FDI treatment than in it is in the Control treatment. Table 4: Summary Statistics Control (Indian) Session 1 Session 2 FDI pval Control FDI (Foreign) (Wilcox) (Indian) (Foreign) pval (t.test) Period 1 giving Period 2 giving Period 3 Info lookup Expected earnings Period 3 giving

25 Period 4 giving N Table 5 below confirms the effect of FDI framing on US students redistribution preferences. The dependent variable is Period 3 giving. We include giving in Period 1 and Period 2 giving in order to control for subject s initial (pre-treatment) altruism. Model (1) includes only subjects in Session 1, Model (2) only subjects in Session (2), and Model (3-5) pools all subjects. Model (6) controls for political affiliation and therefore drops all observation from Session 1. The FDI treatment dummy is positive and significant in all six models, indicating that US subject gives more to the poor when the poor is an Indian with a job opportunity at a foreign firm. Table 5: OLS regression on Period 3 giving (1) (2) (3) (4) (5) (6) Session 1 only Session 2 only Pooled Pooled Pooled Session 2 only Intercept FDI 0.53* 0.4** 0.42* 0.39** 0.33* 0.37* Period 2 giving *** 0.31*** 0.32*** 0.30*** 0.30*** Period 1 giving *** 0.3*** 0.31*** 0.3*** 0.3*** Exp 1 dummy Info purchase -0.13* Expected earnings Demographic controls Age Female 0.46**

26 Democrat R df Why do US subjects think that a poor Indian workers POUM is higher in a domestic (Indian) firm than it is in a foreign firm? Since the subjects did not purchase the necessary information to calculate expected earnings, they must have arrive dupon their beneficiary s POUM through beliefs and biases. What, then, were the bias that drove this result? Separately regressing subjects behavior in Period 3 and 4 by political affiliation provided one possible answer. The dependent variable in Model (1) is the amount of information about the beneficiary s job opportunity that is purchased by the subject. Recall that subjects were offered (and needed) three pieces of information. Democrats purchased 1 piece of information and Republicans purchased slightly less. These differences were, however, not statistically significant. The dependent variable in Model (2) is subjects expected earning of the beneficiary after the introduction of the job opening. While Republicans are unaffected by domestic or foreign framing, Democrats overestimated earnings in Indian firm by a large amount, but gave a much lower estimate if the opening is associated with a foreign firm operating in India. This may be due to the negative coverage of hiring and employment practices of multinational corporations in developing countries. As a result of this bias, redistribution preferences among Democrats show sensitivity to FDI framing (Model (3)). This result is even stronger in Period 4, when subjects has had more time to react to their new expectation of their beneficiary s earnings. Table 6: Info purchase, income expectations: Democrat and Republicans (1)Info purchase (2)Expected earning (3)Period 3 giving (4)Period 4 giving 26

27 Dem Rep Dem Rep Dem Rep Dem Rep Intercept 1.13*** 0.76*** 4.75*** Foreign ** * ** Period 2 giving *** *** 0.47** Period 1 giving * 0.54*** R df Overall our pilot results confirm that benefactors do not invest in costly information when there is a possible change in their beneficiary s well-being. Instead, they react to this potential change by relying on their beliefs and biases to update the deservedness of their beneficiaries. This updated view of their beneficiary is then translated to a change in redistribution preferences. Interestingly, FDI has the opposite effect on redistribution support, than predicted. Is this because liberal in the United States associate multinationals in developing economies with low wages and worker exploitation? Will we find the opposite results for FDI in India? Our next step is to see what belief and biases are used by our Indian subjects to form their redistribution preferences when there is a change that may affect the income opportunities of their low income beneficiaries. Bibliography Ahlquist, John S. (2006) Economic Policy, Institutions, and Capital Flows: Portfolio and Direct Investment Flows in Developing Countries. International Studies Quarterly 50: Aitken, Brian, Ann Harrison, and Robert E. Lipsey Wages and Foreign Ownership: A Comparative Study of Mexico, Venezuela, and the United States. Journal of International 27

28 Economics. 40: Alesina, Alberto and George-Marios Angeletos Fairness and Redistribution. The American Economic Review. 95(4): Alesina, Alberto and Eliana LaFerrara Preferences for Redistribution in the Land of Opportunities. Journal of Public Economics. 89: Alesina, Alberto and Paola Giuliano Preferences for Redistribution. IZA Discussion Paper No Alesina, A.and E. Glaeser Fighting Poverty in the US and Europe: A World of Difference. Oxford: Oxford University Press. J. Andreoni and B.D. Bernheim, Social image and the norm: A theoretical and experimental analysis of audience effects. Econometrica, 77(5): Arbache, Jorge Saba, Andy Dickerson, and Francis Green Trade Liberalization and Wages in Developing Countries. The Economic Journal Beckman, Steve R A Tax and Redistribution Experiment with Subjects that Switch from Risk Aversion to Risk Preference. Social Choice and Welfare. 27(3): Benabou, Roland and Efe A. Ok Mobility as Progressivity: Ranking Income Processes According to Equality of Opportunity. NBER Working Paper Blalock, Garrick and Paul J. Gertler Welfare Gains from Foreign Direct Investment through Technology Transfer to Local Suppliers. Journal of International Economics. 74: Brady, David, Jason Beckfield, and Martin Seeleib-Kaiser Economic Globalization and the Welfare State in Affluent Democracies, American Sociological Review. 70: 921. **Brooks 2010 Burgoon, Brian Globalization and Welfare Compensation: Disentangling the Ties that Bind. International Organization. 55: Checchi, D. and A. Filippin An Experimental Study of the POUM Hypothesis. Research on Economic Inequality. 11: Dana, J., Weber, R.A., Kuang, J.X., Exploiting moral wiggle room: Experiments demonstrating an illusory preference for fairness. Economic Theory. 33, Djankov, Simeon and Bernard Hoekman Foreign Investment and Productivity Growth in Czech Enterprises. The World Bank Economic Review. 14(1):