Are Political Leaders with Professional Background in 1 Business Bad for Climate Mitigation? 2

9 Do political leaders affect the climate mitigation of the nation they govern, and if yes, to which 10 leader characteristics voters who care about climate should pay attention to when they vote? 11 There is abundant literature on how ideology of political parties in power affects climate policy 12 outcomes, but there is nothing similar for individual characteristics of government leaders. 13 This is the first study of its kind, building on a dataset of government leaders of OECD 14 countries for the period 1992-2017, we find that leaders’ professional background is the trait 15 that has the strongest effect. Higher emissions and lower renewable energy deployment are 16 more likely during the tenure of former businesspersons or economists. Teachers and doctors 17 instead are associated with lower emissions and with higher rates of renewable energy 18 deployment. Our results suggest that voters and pressure groups should care about 19 candidates’ professional background, in addition to their party’s ideology.


127
In this piece of research, we empirically test whether leaders' profession, while 128 controlling for other leaders' characteristics, may have an effect on climate mitigation policy 129 and outcomes. And more specifically, the question is whether businesspersons have an impact 130 on climate policy and outcomes, and how. We create a dataset of political leaders' ruling the 131 countries that signed the Kyoto protocol, and examine within-country variations across these 132 leaders' profession and characteristics: gender, family situation, age and years in politics. We 133 do this for a number of countries over the years, while accounting and controlling for 134 contextual differences between leaders, such as years in office, party ideology and whether 135 they govern in coalition or in minority (a proxy for their effective power).

136
Previous datasets in the literature cover long periods of time, but they start before 137 climate policies were introduced, and finish too early for our purposes (early 2000s). We

145
We test associations between the above variations and climate policy and outcomes, 146 measured by the proxies of renewable energy deployment (in terms of installed capacity) and 147 carbon emissions. One is an indicator of effort, and the other of outcomes. We do not expect 148 that the two will move necessarily in the same direction. Until recently, the deployment of 149 renewable energy had not demonstrably displaced fossil fuels (York, 2012); and other policies,

192
A central feature of climate policy is that a leader needs to take a longer-term view 193 towards future generations, a predisposition that one would assume is less likely to be found 194 among leaders with professional backgrounds and social positions that privilege immediate 195 2 These authors use data of California city councils. returns. One plausible expectation is that leaders with professional background in sectors that are trained to prioritize short-term returns, say businesses, will be less likely to act on the 197 climate. In addition, experimental evidence, shows that economists, who share common 198 backgrounds with businesspersons, compared to students enrolled in other university fields, 199 are more prone to free ride (Marwell and Ames, 1981). Thus, it would be not surprising that a 200 political leader who is a businessperson or economist, might be less interested in investing in 201 a "public good" as the climate mitigation, than political leaders with other different 202 backgrounds. 3 Of course, with this reasoning, we do not pretend to rule that businesspersons 203 possess intrinsic personality traits that makes them to be innately less sensitive to the climatic 204 change. Rather, it might also be that they simply are more prone to serve to determinate 205 businesses environments and networks, who may constitute powerful lobbies.  3 These authors ran an experiment intended at maximising the likelihood of free riding. Participants were asked to invest a number of tokens in a collective fund. The share of tokens invested that maximizes the collective benefit was 100%. On average, participants with very heterogeneous backgrounds contributed around 40-50%. Telling participants that the collective good was going to be something non-divisible doubled contributions to about 80%. Only first-year economics graduate students behave very differently, since, on average, they only contributed 20% to the collective fund and a significant number of individuals in this group tried to free ride completely.
For consistency, we include only the twenty-seven countries that were OECD members

224
We determined a leader's profession by looking into his main occupation prior to 225 becoming a professional politician. We classified as "Politician/State Official" those who went 226 directly from school/university to becoming politicians or state officials, or who did not have 227 a clear professional trajectory before becoming politicians (e.g., worked different jobs for a few 228 years). We grouped professions into eight groups: Businesspersons, law-related, college 229 lecturers, politician/civil servants, scientists/science-related, other professions (see Table A1 230 in the Appendix).

231
Our sample includes 156 leaders for a total of 681 leader-year observations (see Table 1). Our 232 average leader is 55 years old, governed for 4 years, and has been in politics for 30 years (Table   233 2). Most leaders are lifetime politicians or civil servants (33%), but there is also a good 234 representation of businesspeople (12%), lawyers (14%), professors (13%) and scientists (15%)

272
higher income countries emitting more and electing specific types of leaders). Reverse 273 causation is a limited concern for our research question. Undoubtedly, there will be a share of

319
Where Z * =(Z * 1, Z * 2, Z * 3) are three leader characteristics that affect our outcome variables (CO2 320 emissions or renewable capacity). Suppose that we are especially interested in measuring the 321 impact of Z * 1 on Y * , and that Z * 1 has an impact on Z * 3 but not on Z * 2, then we can write:

333
According to the citizen-candidate theory, there is a political competition and selection 334 is a game between citizens competing to reach and hold office (Osborne and Slivinski, 1996).

335
This means that many of interactions among individual characteristics we can observe in

360
In

364
The first general conclusion from our results is that leader characteristics matter:  (Table 4).

371
When we compare leaders coming from other professions with businesspersons, we 372 find considerable differences (Table 3). Compared with businesspersons, lawyers and 373 university professors are associated with 6% less emissions, politicians/civil servants 5%,

374
scientists 3%, and school teachers/physicians as much as 16% (though we should treat this 375 last result with caution, as only 7 leaders or 3% of leader years in our sample correspond to 376 teachers/physicians - Table 1). The only category almost as bad as businesspersons are 377 economists with 3% more emissions than the rest of the occupations (Table 4), and no 378 statistically significant difference from businesspersons (Table 3).

379
One might think that the worst performance of businesspersons in climate outcomes is

419
In Table 3, we show the results of the models including all leader's characteristics, while Table   420 4 tests also for the effect of these variables but without controlling for other leader 421 characteristics -this is to check whether some variables that we included in Table 3 (Table 3).

434
Previous literature shows that female politicians are less corrupt or opportunistic 435 (Brollo and Troiano, 2016), are more likely to support foreign aid (Hicks et al., 2016), and also 436 are more prone to invest in infrastructures that are more related to the needs of their own 437 gender (Chattopadhyay and Duflo, 2004). It is also observed that women prefer higher social 438 spending than men (Lott and Kenny, 1999

445
According to our results in Table 3, tenures of female governors are associated with 446 higher levels of carbon emissions by a considerable 8%, though they have no discernible effect 447 on renewable energy capacity. However, as we explained above, gender is highly associated 448 with profession, therefore, it is likely that part of the effect of gender is taken away by 449 profession. Indeed, results reported in Table 4 are somewhat different. We observe that once indicating that a significant effect of women on emissions and renewable capacity operates 454 through its relationship with other variables that we included in Table 3. We know for example 455 that there are fewer women in business and more women that are teachers, professionals in 456 the health sector or civil servants. It is then likely that some of the effect of a leader's profession 457 on the outcome operates indirectly through gender. It makes sense then that taking out variation in terms of professions, which is what we do in Table 3, the effect of women on 459 emissions increases while that in renewables is dampened.

460
Gender does not have to affect policy preferences for innate biological reasons, but 461 through a range of acculturation processes, including training or profession. Table 4, where 462 such factors are not controlled for, gives them a better sense of differences between women 463 and men, as they stand by the time they are leaders. Even so, we see that the negative effect of 464 women on emissions remains statistically significant and considerable (5% more emissions).

465
The small number nonetheless of women leaders in our sample means our finding should be  analysis, though what is striking in our result is not that just women resemble men in climate 482 (in)action, but that they actually perform worse, an intensified Queen-bee effect of a sorts.

483
What requires further study is also why the effect of women on renewable energies would go 484 in the opposite direction to that of carbon emissions. True, as we noted there is no reason why 485 a leader cannot increase during her mandate both renewable energy deployment and carbon 486 emissions, given that the scale of renewable energy is still too small to make a difference. Still, it is not directly clear why would women differ in this to men, assuming that this result is due 488 to a systematic difference.

489
To the best of our knowledge, there is no much evidence about the impact of age, years 490 and politics and years in office on policy outcomes, therefore we cannot build any hypotheses 491 based on previous evidence. However, it seems plausible that in terms of age and experience,

492
we might expect younger politicians to take a longer-term view and hence favour more 493 stringent climate action. Older politicians though might care more about their legacy than 494 short-term political expedience, and they might be more likely to care about the future of their 495 descendants than younger politicians. In this line, we estimate statistically significant impacts 496 (at 5% level) for leader's age and years in politics -however, we find this impact to be generally 497 fairly small.

498
To comment on the impact of age and years in politics, we think that Table 4 is probably 499 in this case a better guide than Table 3, since the age and years in politics naturally co-vary 500 and hence controlling for one while testing for the other, takes away some of the relevant 501 variation of both variables. According to the estimates in Table 4

517
Regarding the variable 'years in office', we are reluctant to draw any generalizing 518 conclusions about seasoned versus 'fresh' politicians. How long a politician stays in power,

519
instead, seems to make a considerable difference, leaders in first term associated with lower 520 emissions, while leaders who have stayed more than 8 years have significantly higher 521 emissions compared to those with shorter mandates (Table 3). One may interpret this as 522 fresher leaders starting with better intentions, an effect which over the years get watered 523 down.

524
Finally, a generational perspective is not observed in the case of parenthood. Parents, 525 that one could expect them to care more about the longer-term impacts of climate change, do 526 not seem to have discernable differences on either carbon emissions or renewable energy from 527 non-parents (Table 3).

575
The main contribution of our research is that it addresses for the firs time the gap in 576 the literature on possible links between political leaders and environmental outcomes, which 577 to the best of our knowledge is virtually inexistent. We think the above results could be better 578 treated as hypotheses for further research, which could mobilize case studies on leaders with 579 interviews, surveys or regression analyses at lower levels of leadership (e.g. regional 580 governors or mayors). Further research could shed light on whether it is the lack of specialized 581 knowledge or lack of training on climate issues in business/economic curricula, or the general 582 profit-first norms cultivated in the business/economics world that drive such differences.

583
Alternatively, it could be proximity or alliances to industrial or fossil fuel interests developed 584 in the professional careers of the leaders that make them reluctant to undertake later action on